Artigo de investigação

RareCode: Uma Estrutura de Aprendizado Profundo Não Supervisionado para Detecção de Anomalias em Imagens Histopatológicas de Câncer Colorretal

9 visualizações

⸱

DOI:

10.3791/72303

⸱

22 de setembro de 2026

 ,  ,  ,  ,  ,  ,  ,  , 
1School of Computer Science and Engineering, Anhui University of Science and Technology, 2School of Medicine, Anhui University of Science and Technology, 3School of Electrical and Information Engineering, Anhui University of Science and Technology, 4The First Hospital, Anhui University of Science and Technology, 5School of Safety Science and Engineering, Anhui University of Science and Technology, 6Department of Pathology, Shenzhen People's Hospital, The Second Clinical Medical College, Jinan University, 7The First Affiliated Hospital, Southern University of Science and Technology, 8Anhui Provincial Academician Workstation of Anhui University of Science and Technology for Autoimmune Disease Diagnostic Technology

Neste artigo

Resumo

Este estudo apresenta o RareCode, um framework de aprendizado profundo não supervisionado que detecta anomalias em imagens histopatológicas de câncer colorretal ao analisar a Raridade na Ativação do Código (CAR), com potencial para auxiliar no pré-rastreamento sem necessidade de dados de treinamento rotulados.

Resumo

A detecção não supervisionada de anomalias em imagens histopatológicas tem sido amplamente explorada por meio de abordagens baseadas em reconstrução que medem erros de reconstrução, mas esses métodos frequentemente falham em capturar variações patológicas sutis em nível semântico devido à heterogeneidade inerente das texturas teciduais. Para superar essa limitação, este estudo apresenta o RareCode, um framework baseado em quantização vetorial que amplia o paradigma de detecção além do reconstrução em nível de pixel, incorporando análise de ativação de código semântico. A inovação central é a pontuação de Raridade na Ativação do Código (CAR), que caracteriza a frequência de ativação de cada entrada do código durante o treinamento exclusivamente com amostras normais e sinaliza ativações infrequentes como indicadores de anomalia no momento da inferência, complementando assim os erros de reconstrução com discriminação em nível semântico. Com base nesse mecanismo CAR em escala única, um módulo de Código Hierárquico Multiescala (MHC) é ainda introduzido, empregando códigos de tamanhos variados com pesos de fusão aprendíveis para capturar padrões patológicos que vão desde estruturas grossas em nível tecidual até detalhes finos em nível celular. Aproveitando o design multiescala, mapas de calor de anomalia hierárquicos são gerados em cada granularidade do código, fornecendo aos patologistas pistas visuais interpretáveis e multidimensionais que indicam tanto a localização quanto o nível estrutural em que as anomalias ocorrem. A validação cruzada com cinco divisões em um conjunto de dados de imagens histopatológicas de câncer colorretal clinicamente anotado, obtido no Hospital de Shenzhen, demonstra que o RareCode alcança uma área sob a curva (AUC) de 96,82%, superando os métodos de referência. A análise por classe mostrou desempenho mais forte na detecção de câncer (AUC = 99,44%, especificidade = 94,21%) do que na detecção de inflamação (AUC = 94,30%, especificidade = 60,44%). Esses resultados sugerem que a análise CAR pode oferecer uma abordagem não supervisionada promissora para detecção de anomalias histopatológicas, potencialmente auxiliando no pré-rastreamento clínico do diagnóstico de câncer colorretal.

Introdução

A triagem histopatológica confiável para câncer colorretal permanece limitada pela anotação de especialistas e revisão manual, especialmente quando lesões inflamatórias e malignidades apresentam morfologia sobreposta1,2. Essas limitações motivam abordagens computacionais não supervisionadas que aprendem com tecido normal sem exigir anotações exaustivas de anomalias3,4.

A urgência clínica do rastreamento histopatológico automatizado é acentuada pelo crescente desequilíbrio entre a demanda diagnóstica e a expertise patológica disponível, com a carga de trabalho diagnóstica por patologista nos EUA aumentando mais de 40% em uma única década, enquanto o número de patologistas continuava a diminuir5. No câncer colorretal (CRC), programas organizados de rastreamento têm sido associados a reduções na mortalidade de 29-68%6, ainda assim a capacidade de revisão patológica manual permanece limitada pela disponibilidade de profissionais. Um sistema automatizado de pré-rastreamento que identifique com confiança casos suspeitos para revisão prioritária poderia ajudar a aliviar esse encargo e melhorar os tempos de resposta diagnóstica. Contudo, a implantação de tais sistemas exige alta sensibilidade para evitar a perda de verdadeiros positivos e especificidade adequada para prevenir alarmes falsos excessivos7.

A detecção não supervisionada de anomalias (UAD) tornou-se cada vez mais relevante na análise de imagens médicas8, pois pode modelar distribuições de tecidos normais sem necessidade de rótulos de treinamento anormais3,9,10. Modelos baseados em reconstrução, incluindo autoencoders, autoencoders variacionais, redes generativas adversariais e variantes com memória aumentada, identificam anomalias por meio de erros de reconstrução, mas decodificadores de alta capacidade ainda podem reconstruir regiões anormais com alta fidelidade11,12,13,14,15,16,17. Métodos baseados em características, como PatchCore e PaDiM, utilizam representações pré-treinadas, embora características aprendidas a partir de imagens naturais possam não capturar plenamente atipias microscópicas em histopatologia18,19,20,21. Modelos fundamentais específicos de patologia e detectores de anomalias baseados em difusão oferecem alternativas poderosas, mas seus requisitos de dados ou custos computacionais podem limitar seu uso direto em triagem de alto rendimento22,23. Métodos recentes automatizados e evolutivos, como EvoAAE24 e MoARNN-AM25, ilustram ainda mais o valor da otimização adaptativa de modelos para detecção de anomalias, embora seus contextos de aplicação difiram da análise de imagens histopatológicas. Em contraste, métodos baseados em quantização vetorial (VQ) impõem restrições de código de livro discretas que podem reduzir o mapeamento de identidade; no entanto, abordagens baseadas em VQ existentes ainda dependem principalmente de erros de reconstrução espacial e subutilizam a informação semântica contida nos padrões de ativação do código de livro26,27.

Embora os métodos mencionados demonstrem desempenho promissor em domínios gerais, sua aplicação a cenários histopatológicos complexos enfrenta desafios específicos. As imagens histopatológicas são caracterizadas por heterogeneidade tecidual complexa em múltiplos níveis estruturais28. Na prática, a análise de imagens normalmente opera em fragmentos locais extraídos de cortes de tecido, e o diagnóstico patológico é inerentemente multiescalar: os patologistas avaliam a morfologia glandular e tecidual em baixa ampliação, enquanto examinam pleomorfismo nuclear e figuras mitóticas em alta ampliação29. Identificamos três limitações principais das abordagens atuais: Primeiro, anomalias patológicas e tecidos normais exibem alta similaridade em características visuais de baixo nível, como estilos de coloração e texturas locais, fazendo com que métodos baseados em reconstrução falhem na detecção de anomalias sutis, já que amostras normais e anômalas podem produzir qualidade de reconstrução semelhante no nível semântico. Segundo, a maioria dos métodos existentes utiliza extração de características em uma única escala, o que dificulta a captura simultânea de características anômalas nos níveis tecidual e celular30,31. Terceiro, embora os métodos convencionais possam gerar mapas de calor em nível de pixel, frequentemente falham em capturar de forma intuitiva os atributos hierárquicos das anomalias, limitando a utilidade desses modelos como auxílios diagnósticos clínicos.

Para enfrentar os desafios acima, propomos o RareCode, um framework UAD que explora padrões de ativação de codebook e representação hierárquica de características em múltiplos níveis de abstração. O framework introduz três componentes principais: (1) Mecanismo de Pontuação de Raridade de Ativação de Codebook (CAR), que calcula pontuações de raridade com base nas frequências de ativação de entradas treinadas exclusivamente em amostras normais, distinguindo amostras normais de anômalas no nível semântico e fornecendo sinais discriminativos complementares aos erros tradicionais de reconstrução espacial. (2) Arquitetura de Fusão de Codebook Hierárquico Multiescalar (MHC), que captura características patológicas em múltiplos níveis de granularidade — desde padrões estruturais grossos até detalhes celulares finos — utilizando codebooks com capacidades variadas, com fusão adaptativa alcançada por meio de pesos aprendíveis. (3) Módulo de Localização Interpretável Hierárquico, que aproveita a arquitetura multiescalar para gerar mapas de calor de anomalias em diferentes níveis de granularidade, fornecendo aos patologistas referências diagnósticas semanticamente interpretáveis em múltiplas escalas.

A hipótese principal deste estudo foi que os perfis de frequência de ativação de código derivados de códigos treinados exclusivamente em imagens histopatológicas normais do cólon e reto codificariam informações semânticas relevantes para anomalias além do erro de reconstrução no nível dos pixels, e que a integração em múltiplas escalas desses sinais complementares melhoraria a discriminação de imagens contendo tecido canceroso ou inflamatório em relação a imagens normais, comparado a métodos representativos de detecção não supervisionada de anomalias (UAD), avaliado principalmente pelo AUC. Para testar essa hipótese, avaliamos o RareCode utilizando validação cruzada de 5 dobras em um conjunto de dados de CCR clinicamente anotado e realizamos análises de ablação e de localização para examinar as contribuições e a interpretabilidade de seus componentes principais.

Protocolo

This study was approved by the Clinical Research Ethics Committee of Shenzhen People's Hospital (Approval No. LL-KY-2025300-01). The requirement for informed consent was waived by the ethics committee because this retrospective study used existing histopathological images and clinical materials without direct patient contact or intervention. All clinical data and histopathological images were de-identified before analysis, and no personally identifiable information was used in this study. All data used in this research were handled in accordance with the ethical standards of the institutional and national research committees.

Framework overview
The overall architecture of the proposed UAD framework, RareCode, is illustrated in Figure 1. The framework employs a parallel dual-branch architecture. For each 512 x 512 input image, non-overlapping 32 x 32 and 64 x 64 patches are extracted as fine- and coarse-scale inputs, respectively. The 64 x 64 patches are resized to 32 x 32 before being passed to the network so that both branches share the same encoder-decoder input size while preserving different receptive fields. Encoders in each branch map the extracted features into a latent space, where the MHC module imposes discretization constraints. Subsequently, decoders reconstruct the image from quantized features to compute spatial-domain reconstruction errors. The CAR mechanism then measures the degree of semantic-level anomaly by profiling codebook activation frequencies. Finally, the model fuses reconstruction scores and CAR scores through a weighted combination to obtain image-level anomaly scores. The RareCode framework is trained end-to-end using only normal samples, requiring no anomaly annotations. The dual-branch design was used to capture both local cellular features and broader glandular structures. The MHC module was used to model these features with codebooks of different capacities.

Encoder-decoder architecture
RareCode constructs structurally independent encoders and decoders for the fine- and coarse-scale branches. Each branch uses the same encoder-decoder design but does not share parameters. The encoder consists of four convolutional blocks, each including a 3 x 3 convolution, batch normalization, ReLU activation, and dropout. The channel width increases from 64 to 128, then to 256, and the final feature map is flattened and projected into a 64-dimensional latent embedding. The decoder mirrors the encoder with a fully connected projection layer followed by four transposed convolution layers, reconstructing each patch to the original input size of the network. Reconstruction error is calculated as the mean squared error between the input and reconstructed patches. By combining this encoder-decoder structure with discrete codebook constraints, RareCode learns compact representations of normal tissue and measures deviations during inference.

Multi-scale hierarchical codebook
To capture pathological features at multiple levels of abstraction—from broad tissue patterns to localized cellular variations—the MHC module employs K discrete codebooks figure-protocol-1 with varying capacities n1, n2, ..., nK. In the final FourScales configuration, each branch contains four codebooks with 64, 128, 256, and 512 entries, respectively, and each codebook entry has a 64-dimensional embedding. For each encoded patch feature, the nearest codebook entry is selected according to Euclidean distance. The quantized outputs from different codebooks are then fused using learnable weights normalized across codebooks. Smaller codebooks, due to stronger compression constraints, are expected to encode coarse-grained prototypical patterns that abstract away local variations, whereas larger codebooks preserve finer-grained features that capture more specific structural characteristics. For each codebook figure-protocol-2, where ek(i) figure-protocol-3 Rd denotes the i-th embedding vector in the k-th codebook, continuous features are mapped to the discrete codebook space via nearest neighbor lookup (Eq. 1 and 2):

figure-protocol-4

figure-protocol-5

To achieve adaptive fusion of multi-scale features, we introduce learnable weights . After softmax normalization, weighted summation is performed over the quantized outputs from each codebook (Eq. 3):

figure-protocol-6

where σ(·) denotes the softmax function, ensuring that weights satisfy Σk σ(wk) = 1. This design enables the model to automatically adjust the contribution ratios of codebooks at different granularities based on the semantic content of the input features.

To optimize codebook learning, we adopt the standard VQ loss function (Eq. 4):

figure-protocol-7

where sg[·] denotes the stop-gradient operation, and β = 0.25 is the commitment loss weight coefficient. The first term encourages codebook vectors to move toward encoder outputs, while the second term encourages encoder outputs to remain consistent with their corresponding codebook vectors. The complete mechanism of multi-scale VQ is illustrated in Figure 2.

Codebook activation rarity scoring
CAR measures semantic-level anomalies based on codebook activation statistics estimated from normal training samples. After model training, all normal training images were passed through the encoder and MHC module without data augmentation. For each branch and each codebook, the assignment index of every patch feature was recorded, and the number of assignments to each codebook entry was counted. The counts were then normalized to obtain the activation-frequency distribution for that codebook (Eq. 5):

figure-protocol-8

During inference, each test image was processed through the same patch extraction, encoding, and nearest-neighbor codebook assignment procedure. For each assigned codebook entry, the rarity score is calculated as the negative logarithm of its activation frequency in the normal training set (Eq. 6):

figure-protocol-9

where ε is a small constant used for numerical stability. A low-frequency codebook entry, therefore, receives a higher rarity score, suggesting that the corresponding patch feature is less consistent with the learned normal-tissue distribution. Patch-level rarity scores were averaged within each image and across the fine- and coarse-scale branches to obtain an image-level rarity response.

To complement activation-frequency rarity, we also compute a percentile-based quantization distance score (Eq. 7):

figure-protocol-10

This score is derived from the Euclidean distance between each encoded feature vector and its nearest codebook entry. The distance distribution is estimated from normal training samples, and test-sample distances are converted into percentile scores. Larger percentile scores indicate that the encoded feature is more difficult to represent using the learned normal codebook prototypes.

The final CAR score combines activation rarity and quantization distance across all codebooks using the learnable codebook weights (Eq. 8):

figure-protocol-11

The same normalized codebook weights used for multi-scale feature fusion are applied to score-level fusion, maintaining consistency between representation learning and anomaly scoring. The resulting CAR score is calculated at the image level and is then combined with the reconstruction score to obtain the final anomaly score. The overall CAR scoring mechanism is illustrated in Figure 3.

Final anomaly score computation
We fuse spatial-domain reconstruction errors with semantic-level CAR scores. Before fusion, both scores are separately normalized using percentile-based min-max normalization to reduce the influence of extreme outliers. The final image-level anomaly score is then defined as (Eq. 9):

figure-protocol-12

where Srecon denotes the normalized mean squared reconstruction error, reflecting sample reconstruction quality in pixel space; SCAR is the CAR score described above, capturing degrees of semantic-level anomaly. The hyperparameter α figure-protocol-13 [0,1] controls the relative weighting between the two components, with optimal values determined on validation sets.

Training objective
The RareCode framework is trained end-to-end using only normal samples. The total training loss comprises the following components (Eq. 10):

figure-protocol-14

Each term is defined as follows: Reconstruction Loss: figure-protocol-15, measuring the mean squared error between input patches p and reconstructed patches D(zq) after quantization. This loss is applied separately to the 32 × 32 patch branch (figure-protocol-16) and 64 × 64 patch branch (figure-protocol-17) to ensure effective learning of features at both spatial scales. 

Dataset description
To evaluate RareCode, experiments were conducted on a clinically annotated CRC histopathological image dataset comprising images of colorectal tissue sections stained with hematoxylin and eosin (H&E), obtained from Shenzhen People's Hospital. All images in this dataset originate from real clinical cases. Images were digitized at 40x magnification with an original resolution of 1024 × 1024 pixels. To ensure annotation reliability and quality, all sample category labels were jointly reviewed and confirmed by two or more senior professional pathologists following a double-blind protocol.

Based on histopathological characteristics, the dataset was categorized into three classes: normal tissue (1,226 images), cancer (1,215 images), and inflammation (1,214 images). In the binary UAD setting, cancer and inflammation samples were treated as anomalous, while normal samples served as the reference distribution. This formulation reflects the intended triage task of separating cases requiring further pathological review from those that are morphologically normal.

All H&E-stained images were resized to 512 × 512 pixels before being used as network input. Stratified 5-fold cross-validation was used for evaluation. In each fold, the normal samples were first divided into training, validation, and held-out test subsets. RareCode was trained only on the training normal samples. The validation subset, which also contained only normal samples, was used for model selection, hyperparameter tuning, and selection of the fusion weight α. The held-out test set, consisting of unseen normal samples and anomalous samples including cancer and inflammation, was used only for final performance evaluation and was not used for model training, hyperparameter selection, or α selection. The representative categories, dataset structure, and validation protocol are shown in Figure 4.

Implementation details
The proposed RareCode framework and comparison baselines were implemented using PyTorch. All experiments were conducted on a workstation equipped with a single NVIDIA GeForce RTX 4060 Ti GPU (16 GB). For network architecture configuration, to capture features at different spatial extents, the dual-branch architecture processes non-overlapping patches at two scales: 32 × 32 pixels (smaller receptive field, capturing local texture patterns) and 64 × 64 pixels (larger receptive field, capturing broader spatial context), respectively. The continuous feature embedding dimension produced by both branch encoders is uniformly set to 64. In the MHC module, we configure four codebooks with different capacities for each branch, with discrete prototype quantities of 64, 128, 256, and 512, respectively.

During optimization, the network was trained end-to-end for 50 epochs with a batch size of 8. We employ the AdamW optimizer for weight updates, with an initial learning rate of 5 × and weight decay of to prevent overfitting. Additionally, to ensure smooth training convergence, a cosine annealing learning rate schedule was used, enabling finer optimization in later training stages. Patch sizes and codebook sizes were selected to balance multi-scale representation and computational cost. The final FourScales configuration was supported by ablation analysis, and α was selected on the validation set before test-set evaluation.

For the decoder-complexity ablation, a complex-decoder variant was implemented based on the FourScales configuration. This variant used the same data split, codebook sizes, training epochs, optimizer, learning rate, validation-based α selection, and evaluation metrics as the FourScales model. The basic decoder was replaced with an EMCAD-style decoder containing multi-scale depthwise convolution blocks and Convolutional Block Attention Modules (CBAM). After each of the first three transposed-convolution upsampling stages, parallel 3 × 3, 5 × 5, and 7 × 7 depthwise convolutions were applied, followed by 1 × 1 pointwise fusion, batch normalization, ReLU activation, a residual connection, and CBAM attention. CBAM included channel attention based on average- and max-pooling descriptors and spatial attention using a 7 × 7 convolution. The final stage used transposed convolution and sigmoid activation to reconstruct 32 × 32 patches.

Resultados

Comparison with baseline methods
To evaluate the RareCode framework, multiple representative UAD methods were selected as baselines, spanning reconstruction-based methods (CAE32, VAE12, SAE33, MAE34, PatchSAE35), memory-augmented reconstruction (MemAE36), knowledge distillation (STFPM37), synthetic anomaly generation (DRAEM38), and pretrained feature extraction (PatchCore18). Baseline selection focused on methods that can be trained or applied under comparable computational budgets and data assumptions (access only to unlabeled normal training samples), thereby ensuring that performance differences reflect methodological contributions rather than disparities in pretraining data scale. To ensure fair comparison, all methods were evaluated using the same 5-fold data splits, preprocessing pipeline, evaluation metrics, and computational environment. Models were trained under the same UAD protocol using only normal training samples, with hyperparameters and thresholds selected on the validation set and final performance assessed only on the held-out test set. Table 1 summarizes 5-fold cross-validation results, and Figure 5 provides multi-metric radar chart comparisons.

As shown in Table 1 and Figure 5, RareCode achieves favorable detection performance and statistical stability on the CRC dataset. In terms of AUC, RareCode (96.82 ± 0.23%) outperforms multiple reconstruction-based baseline models, including MemAE (94.97 ± 0.81%). Statistical analysis confirms that RareCode outperforms all reconstruction-based baselines (paired t-tests, p < 0.05), with the improvement over MemAE reaching statistical significance (p < 0.01). RareCode exhibits low performance variance (standard deviation of 0.23%), indicating consistent cross-fold stability. The limited effectiveness of DRAEM, based on synthetic anomalies, likely reflects the fact that simple texture overlay strategies cannot adequately simulate complex pathological atypia.

Regarding specificity, RareCode achieves 58.24 ± 5.99%, outperforming all comparison methods in reducing false positive rates. This represents approximately a 25-percentage-point improvement over the reconstruction-only baseline (NoCAR, 33.76%), demonstrating the CAR mechanism's contribution to reducing false positives. The lower specificity of STFPM and PatchCore, which rely natural-image-pretrained features, may partially reflect the domain gap between natural and histopathological images. Notably, STFPM and DRAEM exhibit high specificity variance (±33.53% and ±7.09%), with per-fold specificity ranging from near-zero to moderate levels, raising concerns about their deployment reliability.

Per-class detection performance
Per-class analysis was conducted by separately comparing cancer and inflammation samples against normal samples (Table 2). RareCode achieved higher performance for cancer detection (AUC = 99.44 ± 0.12%, recall = 98.13 ± 0.29%, specificity = 94.21 ± 2.22%) than for inflammation detection (AUC = 94.30 ± 0.44%, recall = 96.55 ± 0.78%, specificity = 60.44 ± 4.34%). These findings suggest that RareCode may show stronger discrimination for malignant abnormalities than for inflammatory changes, whereas the inflammation-normal boundary appears to contribute substantially to the reduced overall specificity.

Ablation study
To investigate the contributions of key components within the RareCode framework, ablation experiments were conducted to assess the impact of the CAR mechanism and the MHC module on detection performance. Results are detailed in Table 3. As shown in Figure 6A, adding CAR to the reconstruction-only baseline improved AUC from 92.81% to 95.92%, and multi-scale codebook fusion further increased AUC to 96.82%. Figure 6B shows that CAR also improved specificity, increasing it from 33.76% in the NoCAR baseline to 58.24% in the full FourScales model. These results support the complementary value of codebook activation rarity and multi-scale fusion.

Additionally, a decoder-complexity ablation was conducted using the FourScales configuration. As shown in Table 3, the complex-decoder variant with multi-scale depthwise convolution blocks and Convolutional Block Attention Modules (CBAM) showed lower AUC than the basic FourScales model (95.17 ± 0.44% vs. 96.82 ± 0.23%). This result suggests that, within the present VQ-AE-based RareCode setting, increasing decoder capacity did not improve anomaly detection performance and may reduce the effectiveness of the codebook-constrained representation.

Hierarchical spatial response analysis
To examine the spatial responses generated by RareCode, patch-level reconstruction error and codebook rarity maps were aggregated at the image level and compared between normal and anomalous samples. Energy ratio, Cohen's d, and AUC were used to quantify image-level response separation. Detailed results are presented in Table 4 and Figure 7 and Figure 8.

Results show that codebook rarity scores at all scales exhibit response elevation on anomalous samples (energy ratios > 1). Smaller-capacity codebooks (Codebook 64) achieve stronger localization-level discrimination (78.79% AUC), consistent with the expectation that higher compression rates encourage learning of more abstract prototypical patterns that are more sensitive to deviations from normal tissue archetypes. Reconstruction error alone provides substantial anomaly capture capability (93.31% AUC), while CAR scores offer complementary signals from the semantic frequency dimension.

Qualitative examination of Figure 7 showed that elevated responses in cancer samples overlapped with irregular and crowded glands, nuclear enlargement, hyperchromasia, pseudostratification, and loss of epithelial polarity. In inflammatory samples, stronger responses were observed around distorted crypts and dense inflammatory-cell infiltrates. Lower-capacity codebooks tended to capture broader architectural deviations, whereas higher-capacity codebooks produced more localized cellular-level responses. These findings provide qualitative morphological interpretation but not pixel-level validation.

Fusion parameter analysis
The fusion weight α balances the contributions of spatial reconstruction error and the CAR score to the final anomaly score. Optimal α values for each fold were determined via a grid search (step size 0.05) on the validation sets, with the results presented in Table 5 and  Figure 9. Optimal α values primarily concentrate in the 0.85-0.95 range, with a mean of 0.90 ± 0.05. Under optimal configurations, the model achieves an average AUC of 96.82 ± 0.23% on test sets. Higher optimal weights (approximately 0.90) suggest that CAR scores contribute more to final detection than reconstruction error, while reconstruction error provides auxiliary local constraints. Compared with the α = 0 setting in the sensitivity analysis (AUC = 93.29%), the validation-selected fusion setting improved AUC by approximately 3.53 percentage points.

To summarize the present work presents the RareCode framework for unsupervised anomaly detection in histopathological images, aiming to mitigate the identity-mapping problem encountered by traditional generative models. The CAR mechanism reduces over-reliance on pixel-level reconstruction errors, while the MHC module provides hierarchical feature representations at multiple granularities, enabling complementary anomaly discrimination at different levels of abstraction. Experiments on the CRC dataset demonstrate that RareCode achieves 96.82% AUC, with per-class analysis showing effective cancer detection (AUC = 99.44%) and stronger discrimination for cancer than inflammation, suggesting that inflammation-normal overlap remains a major challenge. Ablation studies confirm that integrating discrete semantic features from VQ with multi-scale tissue morphological representations improves detection performance. RareCode's high recall (98.40%) and moderate specificity (58.24%) suggest its potential as a pre-screening tool for prioritizing suspicious histopathological images; however, its actual effect on pathologist workload will require prospective workflow evaluation.

Data Availability:
The CRC histopathological image dataset used in this study was obtained from Shenzhen People's Hospital under institutional ethics approval (Approval No. LL-KY-2025300-01). Due to patient privacy and institutional data sharing policies, the dataset is not publicly available. Access may be granted upon reasonable request to the corresponding author, subject to institutional approval and data use agreements. The source code for the RareCode framework is publicly available at https://github.com/XL-alg/RareCode.

figure-results-1
Figure 1: Overall architecture and data flow of the RareCode framework. A 512 × 512 H&E-stained histopathological image is divided into non-overlapping 32 × 32 and 64 × 64 patches as fine- and coarse-scale inputs. The two branches encode patches into latent embeddings, apply Multi-scale Hierarchical Codebook (MHC) discretization, and reconstruct patches to compute reconstruction-error scores. In parallel, the Codebook Activation Rarity (CAR) mechanism estimates semantic rarity from codebook activation-frequency profiles learned from normal training samples. Normalized reconstruction and CAR scores are then fused to generate the final image-level anomaly score, while patch-level responses are projected back to the image plane for hierarchical localization. Please click here to view a larger version of this figure.

figure-results-2
Figure 2: Multi-scale vector quantization mechanism. (A) Distance computation between encoder output features and codebook embedding vectors. (B) Nearest neighbor selection and weighted fusion across codebooks with capacities of 64, 128, 256, and 512. (C) Reconstruction from quantized features via transposed convolutional decoder. Please click here to view a larger version of this figure.

figure-results-3
Figure 3: CAR scoring mechanism. This figure illustrates four stages: Stage 1: activation frequency statistics are computed solely on normal training samples. Stage 2: test samples obtain activation indices and distances through encoder and vector quantization. Stage 3: rarity scores and distance scores are computed based on training-set frequency profiles. Stage 4: scores from each codebook scale are fused through learnable weights to produce the final CAR score. Please click here to view a larger version of this figure.

figure-results-4
Figure 4: CRC dataset composition and experimental protocol. (A–C) Representative H&E-stained histopathological images of normal colorectal tissue, colorectal cancer, and colorectal inflammation. (D) 5-fold cross-validation protocol for UAD, in which normal samples were used for training and validation, while held-out normal and anomalous samples were used for testing. Images were digitized at 40x magnification and processed into 32 x 32 and 64 x 64 patches. Green, light green, and orange indicate training, validation, and test sets, respectively. Please click here to view a larger version of this figure.

figure-results-5
Figure 5: Multi-metric comparison radar chart of anomaly detection methods. Radar chart comparing RareCode against baseline methods across AUC, average precision (AP), F1-score, precision, recall, and precision at 90% recall (P@R90). All values represent mean performance across 5-fold cross-validation. Please click here to view a larger version of this figure.

figure-results-6
Figure 6: Incremental contribution analysis of ablation study components. (A) AUC performance comparison showing the progressive improvement from reconstruction-only baseline (NoCAR) through single-codebook to multi-scale configurations. (B) Specificity performance comparison demonstrating the CAR mechanism's effect on false positive reduction. All error bars represent standard deviation (SD) across 5-fold cross-validation. Please click here to view a larger version of this figure.

figure-results-7
Figure 7: Multi-scale codebook hierarchical localization heatmaps. Representative examples are shown for (A) normal, (B) cancer, and (C) inflammation samples. Each row includes the original image, overlay, reconstruction-error map, rarity-score maps from codebooks of different capacities (64, 128, 256, and 512), and the final fused localization map. Smaller-capacity codebooks highlight coarse-grained patterns through stronger compression abstraction, whereas larger-capacity codebooks preserve finer structural details. High-response regions qualitatively correspond to cancer- or inflammation-related histopathological features, but were not validated with pixel-level expert annotations. Please click here to view a larger version of this figure.

figure-results-8
Figure 8: Distribution histograms of different score types. Histograms compare score distributions between normal and anomalous samples for (A) reconstruction error, (B) combined CAR score, (C) Codebook 64, (D) Codebook 128, (E) Codebook 256, and (F) Codebook 512. Green and red histograms indicate normal and anomalous samples, respectively. Please click here to view a larger version of this figure.

figure-results-9
Figure 9: Alpha parameter sensitivity analysis. (A) Validation-set AUC used for α selection. (B) Test-set AUC under different α values. The validation-selected α values ranged from 0.85 to 0.95, with a mean of 0.90 ± 0.05, consistent with Table 5. AUC variation remained below 0.5% when α varied within the [0.70, 1.00] range, indicating stable performance across a wide parameter range. Please click here to view a larger version of this figure.

MethodAUC (%)AP (%)F1 (%)Precision (%)Recall (%)Specificity (%)P@R90 (%)
CAE90.37 ± 0.9498.62 ± 0.1895.34 ± 0.1591.64 ± 0.4099.35 ± 0.2228.14 ± 3.8895.66 ± 0.32
SAE90.58 ± 0.9498.66 ± 0.1895.34 ± 0.1591.67 ± 0.4099.32 ± 0.2428.46 ± 3.9795.71 ± 0.30
VAE93.60 ± 0.8299.13 ± 0.1295.86 ± 0.1992.81 ± 0.4999.12 ± 0.2939.07 ± 4.7096.94 ± 0.43
MAE91.65 ± 2.7598.76 ± 0.5095.61 ± 0.6092.87 ± 1.5698.55 ± 0.5539.74 ± 14.3296.55 ± 0.97
MemAE94.97 ± 0.8199.35 ± 0.1195.78 ± 0.3392.82 ± 1.0798.95 ± 0.6039.15 ± 9.9997.57 ± 0.33
PatchSAE94.86 ± 0.3699.34 ± 0.0595.39 ± 0.1392.32 ± 0.2798.67 ± 0.1234.91 ± 2.5798.13 ± 0.24
STFPM90.23 ± 3.0598.70 ± 0.4494.32 ± 0.2191.90 ± 3.5197.14 ± 3.3829.82 ± 33.5395.93 ± 1.96
DRAEM76.34 ± 2.8295.34 ± 0.9094.19 ± 0.0789.39 ± 0.6599.56 ± 0.656.19 ± 7.0992.69 ± 0.57
PatchCore74.64 ± 1.8294.76 ± 0.5594.73 ± 0.0590.52 ± 0.1599.36 ± 0.1217.49 ± 1.5492.54 ± 0.15
RareCode96.82 ± 0.2399.59 ± 0.0396.63 ± 0.1594.93 ± 0.6798.40 ± 0.4858.24 ± 5.9998.80 ± 0.10

Table 1: Comparison results with baseline methods (5-fold cross-validation). Detection performance of RareCode and nine baseline UAD methods on the CRC dataset. Metrics include AUC, average precision (AP), recall, specificity, F1-score, precision at 90% recall (P@R90), and precision. All values represent mean ± SD across 5-fold cross-validation. Bold values indicate the best performance for each metric.

CategoryAUC (%)AP (%)F1 (%)Recall (%)Specificity (%)
Cancer99.44 ± 0.1299.86 ± 0.0398.34 ± 0.1798.13 ± 0.2994.21 ± 2.22
Inflammation94.30 ± 0.4498.48 ± 0.1293.44 ± 0.2996.55 ± 0.7860.44 ± 4.34

Table 2: Per-class detection performance across anomaly subtypes. Separate evaluation of RareCode's detection performance for cancer and inflammation samples against normal samples. Per-class metrics, including AUC, AP, F1-score, recall, and specificity, were computed using class-specific optimal thresholds.

VariantCodebook ConfigCARAUC (%)AP (%)F1 (%)Specificity (%)P@R90 (%)
NoCAR  figure-results-10  figure-results-1192.81 ± 0.1299.05 ± 0.0295.30 ± 0.0833.76 ± 3.8296.72 ± 0.12
SingleCodebook[256]  figure-results-1295.92 ± 0.4099.47 ± 0.0596.25 ± 0.1455.37 ± 6.5198.31 ± 0.43
TwoScales[128, 512]  figure-results-1396.57 ± 0.2799.56 ± 0.0496.45 ± 0.1257.75 ± 4.1498.75 ± 0.12
ThreeScales[128, 256, 512]  figure-results-1496.36 ± 0.3799.53 ± 0.0596.52 ± 0.1757.99 ± 4.6598.60 ± 0.23
FourScales[64, 128, 256, 512]  figure-results-1596.82 ± 0.2399.59 ± 0.0396.63 ± 0.1558.24 ± 5.9998.80 ± 0.10
FourScales + complex decoder[64, 128, 256, 512]  figure-results-1695.17 ± 0.4499.39 ± 0.0695.32 ± 0.2053.83 ± 21.4598.40 ± 0.37

Table 3: Ablation study results. Performance comparison of RareCode configurations with progressive addition of components: reconstruction-only baseline (NoCAR), single-codebook CAR (SingleCodebook), and multi-scale configurations (TwoScales, ThreeScales, FourScales). An additional decoder-complexity ablation using the FourScales configuration is also included. Metrics include AUC, average precision (AP), F1-score, specificity, and precision at 90% recall (P@R90).

Score TypeEnergy RatioCohen's dAUC (%)
Reconstruction Error2.6232.00693.31
Combined CAR1.0781.00274.85
Codebook 641.1091.20778.79
Codebook 1281.0851.06776.19
Codebook 2561.0650.91672.80
Codebook 5121.0640.83370.38

Table 4: Image-level analysis of localization-derived responses. Image-averaged reconstruction error and codebook rarity responses were compared between normal and anomalous samples using the energy ratio, Cohen's d, and AUC.

FoldOptimal αVal AUC (%)Test AUC (%)
10.9596.2296.91
20.996.3896.39
30.8596.7196.82
40.8596.2296.93
50.9596.7297.05
Mean0.90 ± 0.0596.45 ± 0.2396.82 ± 0.23

Table 5: Optimal alpha values across folds. Optimal fusion weight alpha determined via grid search (step size 0.05) on validation sets for each cross-validation fold, with mean and SD statistics.

Discussão

Os resultados sugerem que os padrões de ativação do codebook podem fornecer informações complementares aos erros de reconstrução espacial. A melhoria do NoCAR até o modelo completo FourScales apoia a contribuição potencial da pontuação CAR, enquanto os pesos de fusão selecionados indicam que a raridade semântica pode desempenhar um papel importante na pontuação final de anomalia. Uma explicação possível é que as anormalidades relacionadas ao CCR podem abranger diferentes escalas morfológicas, desde atipia nuclear até a desorganização arquitetural glandular, e, portanto, podem se beneficiar de representações de características em múltiplas granularidades14. O modelo atual captura diferentes extensões espaciais utilizando fragmentos de 32 × 32 e 64 × 64 no mesmo nível de ampliação; a integração futura com estruturas de processamento de imagens de lâmina inteira (WSI) pode apoiar ainda mais a análise hierárquica em nível de lâmina39.

Clinicamente, o RareCode pode ser mais bem considerado como uma ferramenta de triagem prévia, e não como um sistema diagnóstico autônomo. Sua alta sensibilidade (98,40%) sugere um risco relativamente baixo de omitir casos anômalos40, enquanto sua especificidade moderada (58,24%) provavelmente reflete a dificuldade em distinguir alterações inflamatórias de tecido normal em um contexto não supervisionado. A especificidade mais alta específica para câncer (94,21%) sugere ainda que alarmes falsos podem ser menos frequentes para o subtipo clinicamente mais crítico. Essas características de desempenho apoiam o uso potencial do RareCode como ferramenta de pré-triagem para priorizar casos que exigem avaliação especializada e agilizar a revisão patológica5,6.

As descobertas por classe podem reforçar ainda mais o potencial papel do RareCode no triagem. O desempenho relativamente superior para câncer em comparação com inflamação pode refletir os desvios arquitetônicos e citológicos mais acentuados frequentemente observados em tecido maligno, incluindo perda de polaridade, pleomorfismo nuclear e desmoplasia estromal.14,39. Em contraste, as alterações inflamatórias podem sobrepor-se mais estreitamente à variação tecidual benigna, o que poderia explicar parcialmente a menor especificidade específica para inflamação. Portanto, a especificidade geral deve ser interpretada com cautela, considerando a detecção de câncer como um caso de uso clinicamente importante, e não como um diagnóstico autônomo definitivo.

Os possíveis modos de falha do RareCode devem ser interpretados no contexto do seu objetivo de aprendizado monoclassificado. Como o modelo aprende padrões de tecido normal em vez de categorias específicas de doenças, alterações não neoplásicas, como epitélio regenerativo, fibrose, necrose ou inflamação acentuada, também podem receber altas pontuações de anomalia. Por outro lado, displasia sutil ou carcinoma bem diferenciado com arquitetura glandular quase normal pode gerar respostas mais fracas. Fatores técnicos, incluindo variação na coloração, dobras no tecido, artefatos de corte, desfoque e compressão da imagem, também podem influenciar os erros de reconstrução e as frequências de ativação do código. Essas considerações indicam que a tradução clínica exigiria controle de qualidade da imagem, harmonização da coloração, calibração com reconhecimento do scanner e validação multicêntrica39,41.

Escopo das comparações de referência. A avaliação experimental deste estudo concentra-se em métodos que operam sob suposições comparáveis de dados e computação — especificamente, métodos que podem ser treinados de ponta a ponta utilizando apenas pequenas amostras normais não rotuladas, sem necessidade de corpora externos de pré-treinamento. Esse escopo abrange os principais paradigmas na detecção não supervisionada de anomalias: baseados em reconstrução (CAE, VAE, SAE, MAE, PatchSAE), com memória aumentada (MemAE), por distilação de conhecimento (STFPM), por aumento sintético (DRAEM) e por correspondência de características pré-treinadas (PatchCore). Reconhecemos que duas categorias de métodos não foram incluídas como referências diretas. Primeiro, modelos fundamentais específicos de patologia (UNI, CONCH, CTransPath) utilizam pré-treinamento em milhões de imagens de patologia curadas, e suas vantagens de desempenho são atribuídas principalmente à escala e diversidade dos dados de pré-treinamento, e não ao mecanismo de detecção de anomalias; uma comparação direta, portanto, confundiria as contribuições dos dados de pré-treinamento com as contribuições da metodologia de detecção. O desempenho subótimo de STFPM e PatchCore — ambos utilizando estruturas pré-treinadas no ImageNet — ilustra empiricamente o impacto da lacuna de domínio quando características pré-treinadas de uso geral são aplicadas a cenários histopatológicos, sugerindo que o pré-treinamento específico de patologia pode superar essa lacuna. Segundo, métodos de detecção de anomalias baseados em difusão, embora promissores, impõem uma sobrecarga computacional substancialmente maior durante a inferência (geralmente exigindo centenas de etapas iterativas de remoção de ruído), limitando sua aplicabilidade em fluxos de trabalho clínicos de triagem de alto rendimento, onde a eficiência de processamento é um requisito prático. Trabalhos futuros investigarão se a incorporação de codificadores pré-treinados específicos de patologia na arquitetura RareCode — substituindo o codificador atual treinado de ponta a ponta — pode melhorar ainda mais o desempenho de detecção, mantendo as vantagens de interpretabilidade do mecanismo CAR.

Várias limitações devem ser reconhecidas. Primeiro, a validação foi limitada a um conjunto de dados de CCR de um único centro, sendo necessária uma avaliação multicêntrica adicional entre instituições, scanners e protocolos de coloração. Segundo, embora a especificidade específica para câncer tenha sido encorajadora, a especificidade geral permaneceu afetada pelo limite entre inflamação e tecido normal. Terceiro, anotações especializadas em nível de pixel não estavam disponíveis; portanto, o coeficiente de Dice e o índice de sobreposição (IoU) não puderam ser calculados, e a avaliação de localização foi limitada à visualização qualitativa e à análise em nível de imagem das respostas espaciais agregadas. Quarto, estudos formais com leitores ainda são necessários para determinar se mapas de calor hierárquicos podem melhorar a precisão diagnóstica ou a eficiência da revisão. Por fim, o RareCode atualmente realiza análise em nível de fragmento e de imagem, e a implantação em nível de WSI exigiria integração com estruturas adicionais de processamento39,41,42.

Trabalhos futuros devem avaliar o RareCode em bancos de dados públicos e coortes multicêntricas após a integração com estruturas de processamento de WSI. Estudos prospectivos com leitores podem ajudar a avaliar seu impacto potencial na precisão diagnóstica, tempo de revisão e concordância interobservador42. Outras direções incluem a melhoria da eficiência computacional, a extensão do modelo para subtipagem multiclasse de anomalias e a integração de aprendizado baseado em múltiplas instâncias para diagnóstico em nível de lâmina28,39.

Divulgações

Os autores declaram que não possuem interesses financeiros concorrentes conhecidos ou relações pessoais que poderiam ter parecido influenciar o trabalho descrito neste artigo. Durante a elaboração deste trabalho, utilizamos o ChatGPT para aprimorar a linguagem e a legibilidade do manuscrito e para auxiliar na geração de código de visualização de dados. Após utilizar esta ferramenta, revisamos e editamos o conteúdo conforme necessário e assumimos total responsabilidade pelo artigo publicado.

Agradecimentos

Esta pesquisa foi apoiada pela Estação de Trabalho do Acadêmico Yong Dai (Tecnologia Diagnóstica para Doenças Autoimunes) (Ciência e Tecnologia Wan Ke [2023] nº 317), pelo Projeto do Fundo de Inovação para Pós-Graduandos do Centro Nacional de Pesquisa de Hefei para Centro de Pesquisa Conjunta em Ciências da Saúde e Tecnologia em Medicina Ocupacional e Saúde Fundo Aberto (nº OMH-2023-04) e pelo Projeto de Pesquisa Clínica e Traducional da Província de Anhui (nº 202427610020132).

Materiais

Lista de materiais utilizados neste artigo
NomeEmpresaNúmero de catálogoComentários
GPU GeForce RTX 4060 TiNVIDIAN/A16 GB de VRAM; GPU único utilizado em todos os experimentos (RRID:SCR_022858)
MatplotlibEquipe de Desenvolvimento do MatplotlibVersão 3.8.4Visualização de dados e geração de figuras (RRID:SCR_008624)
NumPyDesenvolvedores do NumPyVersão 1.26.4Computação numérica e operações com arrays (RRID:SCR_008633)
pandasEquipe de Desenvolvimento do pandasVersão 2.2.3Organização de dados e processamento de resultados tabulares (RRID:SCR_018214)
PillowEquipe de Colaboradores da Biblioteca Python de Imagem / PillowVersão 10.3.0Carregamento e pré-processamento de imagens
PythonPython Software FoundationVersão 3.12.3Linguagem de programação utilizada para desenvolvimento de modelos e análise de dados (RRID:SCR_008394)
PyTorchMeta Platforms, Inc.Versão 2.7.0+cu118Framework de aprendizado profundo utilizado para implementação e treinamento do modelo (RRID:SCR_018536)
scikit-learnDesenvolvedores do scikit-learnVersão 1.4.2Validação cruzada, divisão treino-teste e métricas de avaliação (RRID:SCR_002577)
SciPyDesenvolvedores do SciPyVersão 1.13.1Análise estatística (RRID:SCR_008058)
torchvisionProjeto PyTorchVersão 0.22.0+cu118Pré-processamento de imagens e transformação de dados
tqdmDesenvolvedores do tqdmVersão 4.66.4Monitoramento do progresso durante treinamento e avaliação

Referências

  1. Fan J, Sun Q, Di Y, et al. DIPathMamba: A domain-incremental weakly supervised state space model for pathology image segmentation. Med Image Anal. 2025;103:103563.
  2. Nan T, Zheng S, Qiao S, et al. Deep learning quantifies pathologists' visual patterns for whole slide image diagnosis. Nat Commun. 2025;16(1):5493.
  3. Lagogiannis I, Meissen F, Kaissis G, et al. Unsupervised pathology detection: a deep dive into the state of the art. IEEE Trans Med Imaging. 2024;43(1):241-252.
  4. Tian Y, Liu F, Pang G, et al. Self-supervised pseudo multiclass pre-training for unsupervised anomaly detection and segmentation in medical images. Med Image Anal. 2023;90:102930.
  5. Metter DM, Colgan TJ, Leung ST, et al. Trends in the US and Canadian pathologist workforces from 2007 to 2017. JAMA Netw Open. 2019;2(5):e194337.
  6. Shaukat A, Levin TR. Current and future colorectal cancer screening strategies. Nat Rev Gastroenterol Hepatol. 2022;19(8):521-531.
  7. Ancker JS, Edwards A, Nosal S, et al. Effects of workload, work complexity, and repeated alerts on alert fatigue in a clinical decision support system. BMC Med Inform Decis Mak. 2017;17(1):36.
  8. Litjens G, Kooi T, Ehteshami Bejnordi B, et al. A survey on deep learning in medical image analysis. Med Image Anal. 2017;42:60-88.
  9. Stepec D, Skocaj D. Unsupervised detection of cancerous regions in histology imagery using image-to-image translation. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops. 2021:3785-3792.
  10. Li Y, Lao Q, Kang Q, et al. Self-supervised anomaly detection, staging and segmentation for retinal images. Med Image Anal. 2023;87:102805.
  11. Hinton GE, Salakhutdinov RR. Reducing the dimensionality of data with neural networks. Science. 2006;313(5786):504-507.
  12. Kingma DP, Welling M. Auto-encoding variational Bayes. In: Proceedings of the 2nd International Conference on Learning Representations (ICLR). 2014.
  13. Goodfellow IJ, Pouget-Abadie J, Mirza M, et al. Generative adversarial nets. In: Advances in Neural Information Processing Systems. 2014;27:2672-2680.
  14. Lu S, Zhang W, Zhao H, et al. Anomaly detection for medical images using heterogeneous auto-encoder. IEEE Trans Image Process. 2024;33:2770-2782.
  15. Yang Z, Zhang T, Soltani Bozchalooi I, et al. Memory-augmented generative adversarial networks for anomaly detection. IEEE Trans Neural Netw Learn Syst. 2022;33(6):2324-2334.
  16. Huyan N, Quan D, Zhang X, et al. Unsupervised outlier detection using memory and contrastive learning. IEEE Trans Image Process. 2022;31:6440-6454.
  17. Jézéquel L, Beaudet J, Histace A, et al. Unified anomaly detection via multi-scale contrasted memory. IEEE Trans Image Process. 2026;35:2802-2815.
  18. Roth K, Pemula L, Zepeda J, et al. Towards total recall in industrial anomaly detection. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. 2022:14298-14308.
  19. Defard T, Setkov A, Loesch A, et al. PaDiM: a patch distribution modeling framework for anomaly detection and localization. In: International Conference on Pattern Recognition. Cham: Springer; 2021:475-489.
  20. Zingman I, Stierstorfer B, Lempp C, et al. Learning image representations for anomaly detection: application to discovery of histological alterations in drug development. Med Image Anal. 2024;92:103067.
  21. Jiang Y, Cao Y, Shen W. Prototypical learning guided context-aware segmentation network for few-shot anomaly detection. IEEE Trans Neural Netw Learn Syst. 2025;36(7):12016-12026.
  22. Wang X, Zhao J, Marostica E, et al. A pathology foundation model for cancer diagnosis and prognosis prediction. Nature. 2024;634(8035):970-978.
  23. Frotscher A, Kapoor J, Wolfers T, et al. Unsupervised anomaly detection in medical imaging using aggregated normative diffusion. Med Image Anal. 2026;109:103895.
  24. Zeng GQ, Yang YW, Lu KD, et al. Evolutionary adversarial autoencoder for unsupervised anomaly detection of industrial Internet of Things. IEEE Trans Reliab. 2025;74(3):3454-3468.
  25. Lu KD, Zhang BX, Xu Y, et al. MoARNN-AM: multi-objective automated recurrent neural network with attention mechanism for cyber-attack detection of UAV. IEEE Trans Consum Electron. 2026;72(1):1738-1749.
  26. van den Oord A, Vinyals O, Kavukcuoglu K. Neural discrete representation learning. In: Advances in Neural Information Processing Systems. 2017;30:6306-6315.
  27. Ghafourian A, Shui H, Upadhyay D, et al. Targeted collapse regularized autoencoder for anomaly detection: black hole at the center. IEEE Trans Neural Netw Learn Syst. 2025;36(6):10348-10358.
  28. Gao Z, Mao A, Dong Y, et al. SMMILe enables accurate spatial quantification in digital pathology using multiple-instance learning. Nat Cancer. 2025;6(12):2025-2041.
  29. Wang X, Liu H, Zhang Y, et al. Joint modeling histology and molecular markers for cancer classification. Med Image Anal. 2025;102:103505.
  30. Jin H, Shen J, Cui L, et al. Dynamic graph based weakly supervised deep hashing for whole slide image classification and retrieval. Med Image Anal. 2025;101:103468.
  31. Schmitz R, Madesta F, Nielsen M, et al. Multi-scale fully convolutional neural networks for histopathology image segmentation: from nuclear aberrations to the global tissue architecture. Med Image Anal. 2021;70:101996.
  32. Masci J, Meier U, Ciresan D, et al. Stacked convolutional auto-encoders for hierarchical feature extraction. In: International Conference on Artificial Neural Networks. Berlin: Springer; 2011:52-59.
  33. Ng A. Sparse autoencoder. CS294A Lecture Notes. 2011;72:1-19.
  34. He K, Chen X, Xie S, et al. Masked autoencoders are scalable vision learners. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. 2022:16000-16009.
  35. Lim H, Choi J, Choo J, et al. Sparse autoencoders reveal selective remapping of visual concepts during adaptation. In: Proceedings of the 13th International Conference on Learning Representations (ICLR). 2025.
  36. Gong D, Liu L, Le V, et al. Memorizing normality to detect anomaly: memory-augmented deep autoencoder for unsupervised anomaly detection. In: Proceedings of the IEEE/CVF International Conference on Computer Vision. 2019:1705-1714.
  37. Wang G, Han S, Ding E, et al. Student-teacher feature pyramid matching for anomaly detection. In: Proceedings of the 32nd British Machine Vision Conference (BMVC). 2021.
  38. Zavrtanik V, Kristan M, Skocaj D. DRAEM: a discriminatively trained reconstruction embedding for surface anomaly detection. In: Proceedings of the IEEE/CVF International Conference on Computer Vision. 2021:8330-8339.
  39. Niazi MKK, Parwani AV, Gurcan MN. Digital pathology and artificial intelligence. Lancet Oncol. 2019;20(5):e253-e261.
  40. Raab SS, Grzybicki DM, Janosky JE, et al. Clinical impact and frequency of anatomic pathology errors in cancer diagnoses. Cancer. 2005;104(10):2205-2213.
  41. Schömig-Markiefka B, Pryalukhin A, Hulla W, et al. Quality control stress test for deep learning-based diagnostic model in digital pathology. Mod Pathol. 2021;34(12):2098-2108.
  42. Steiner DF, MacDonald R, Liu Y, et al. Impact of deep learning assistance on the histopathologic review of lymph nodes for metastatic breast cancer. Am J Surg Pathol. 2018;42(12):1636-1646.

Reimpressões e permissões

Etiquetas

Aprendizado Não SupervisionadoAtivação de CodebookQuantização VetorialCodebook MultiescalaDetecção de CâncerAnálise Semântica
Fale conosco
Recomendar à biblioteca

Privacidade

Termos de Uso

Políticas

Copyright © 2026 MyJoVE Corporation. Todos os direitos reservados.