radiomic-feature-selection-using-gradient-loss-deep-neural-network

Radiomic Feature Selection Using Gradient Loss of Deep Neural Network for Lung Cancer Stage Detection

Radiomics enables the extraction of quantitative imaging biomarkers from medical images and has become an important tool for computer-aided cancer ...

Bahria University

Lung nodule segmentation in CT images and its subsequent volume analysis can help determine the malignancy status of a lung nodule. While several efficient segmentation schemes have been proposed, only a few studies evaluated the segmentation's performance for large nodules. In this research, we contribute a semi-automatic system which is capable of performing robust 3-D segmentations on both small and large nodules with good accuracy. The target CT volume is de-noised with an anisotropic diffusion filter and a region of interest is selected around the target nodule on a reference slice. The proposed model performs nodule segmentation by incorporating a mean intensity based threshold in Geodesic Active Contour model in level sets. We also devise an adaptive technique using image intensity histogram to estimate the desired mean intensity of the nodule. The proposed system is validated on both lung nodules and phantoms collected from publicly available diverse databases. Quantitative and visual comparative analysis of the proposed work with the Chan-Vese algorithm and statistic active contour model of 3D Slicer platform is also presented. The resulting mean spatial overlap between segmented nodules and reference nodules is 0.855, the mean volume bias is 0.10±0.2 ml and the algorithm repeatability is 0.060 ml. The achieved results suggest that the proposed method can be used for volume estimations of small as well as large-sized nodules.

3-D segmentation of lung nodules using hybrid level sets.

Radiomic features based classifiers and neural networks have shown promising results in tumor classification. The classification performance can be further improved greatly by exploring and incorporating the discriminative features towards cancer into mathematical models. In this research work, we have developed two radiomics driven likelihood models in Computed Tomography(CT) images to classify lung, colon, head and neck cancer. Initially, two diagnostic radiomic signatures were derived by extracting 105 3-D features from 200 lung nodules and by selecting the features with higher average scores from several supervised as well as unsupervised feature ranking algorithms. The signatures obtained from both the ranking approaches were integrated into two mathematical likelihood functions for tumor classification. Validation of the likelihood functions was performed on 265 public data sets of lung, colon, head and neck cancer with high classification rate. The achieved results show robustness of the models and suggest that diagnostic mathematical functions using general tumor phenotype can be successfully developed for cancer diagnosis.

Radiomics based likelihood functions for cancer diagnosis.

Cancer survival time prediction using Deep Learning (DL) has been an emerging area of research. However, non-availability of large-sized annotated medical imaging databases affects the training performance of DL models leading to their arguable usage in many clinical applications. In this research work, a neural network model is customized for small sample space to avoid data over-fitting for DL training. A set of prognostic radiomic features is selected through an iterative process using average of multiple dropouts which results in back-propagated gradients with low variance, thus increasing the network learning capability, reliable feature selection and better training over a small database. The proposed classifier is further compared with erasing feature selection method proposed in the literature for improved network training and with other well-known classifiers on small sample size. Achieved results which were statistically validated show efficient and improved classification of cancer survival time into three intervals of 6 months, between 6 months up to 2 years, and above 2 years; and has the potential to aid health care professionals in lung tumor evaluation for timely treatment and patient care.

Article	Year
3-D segmentation of lung nodules using hybrid level sets. Computers in biology and medicine\| PubMed ID: 29631230	2018
Radiomics based likelihood functions for cancer diagnosis. Scientific reports\| PubMed ID: 31263186	2019
A deep learning-based cancer survival time classifier for small datasets. Computers in biology and medicine\| PubMed ID: 37150085	2023

Hina Shakir

Biography

JoVE Journal Publications

Other Publications