Integration of Animal Behavioral Assessment and Convolutional Neural Network to Study Wasabi-Alcohol Taste-Smell Interaction

We want to see if smelling alcoholic drinks could ease wasabi's burn. So we developed a method that involved animal behavioral study and deep learning to explore this taste-smell interaction. By using our specialized cage design for behavioral data acquisition and a robust deep learning model, we found that 29%of the images depicting co-treatment of wasabi and alcohol were classified as wasabi-negative, liquid-positive, suggesting alcohol may counteract wasabi's effects.

Our study prompts the question of, why alcohol can suppress the discomfort caused by wasabi? Understanding the underlying physiological mechanism driving this interaction will be a focus of future research. To begin, construct two opaque side walls by stacking bricks to a height of 54 millimeters.

Embed the acrylic plates into the acrylonitrile butadiene styrene-based cage. Prepare a chows chamber constructed of five 90-millimeter by 50-millimeter transparent acrylic plates with a thickness of two millimeters. Next, prepare a transparent acrylic plate as a chow introduction plate and place it in the chows chamber.

After conducting behavioral experiments using a nine-to 10-week-old mouse model, disassemble all the acrylic and acrylonitrile butadiene styrene plates. Rinse the plates with ultrapure water at least three times. And then dry them using paper towels.

Next, spray them with 75%ethanol, followed by cleaning them with lens paper. Finally, allow them to air dry for at least 15 minutes. Now, weigh the mice and record their body weights before each replication of the behavioral experiment.

To prepare a mixture of wasabi and peanut butter, weigh one gram of commercial wasabi and 4.5 grams of peanut butter. Mix them in a zipped plastic bag. Weigh and provide either two 0.5-gram pastes of peanut butter or a mixture of wasabi and peanut butter on the chow introduction plate.

Place the prepared chow introduction plate in the chows chamber. Then, fill the groove underneath with 30 milliliters of liquid, either pure water or liquor, to facilitate concurrent inhalation. Begin recording using the cameras on two smartphones placed on the phone stands at each terminal.

Carefully place two trained mouse litter mates into the designed animal behavioral study platform from the top and promptly secure the cage with the top plate. To begin, record the behavior of mice during the experience of wasabi-induced nociception while sniffing liquor in a specifically designed cage. Using the provided Jupyter Notebook named Step1_ExtractingAndSavingVideoFrameImages.

ipynb, export a series of video frame images from the collected video clips to generate a dataset for model training. Only select the images with at least one mouse consuming the provided paste. To perform data augmentation, implement the script provided in the Jupyter Notebook named Step2_DataAugmentation.

ipynb to horizontally flip the generated images. Reserve the image data from each second replicate for external independent CNN model validation. Use the images from each first and third replicate for internal model training and testing.

To pre-process the image data used in CNN modeling, run the script in the Jupyter Notebook named Step3_CNNmodeling_TrainTest. ipynb, including image resizing, black color conversion, and image signal normalization. Split the training materials into internal training and testing datasets in an eight-to-two fashion randomly.

Then, initialize the CNN architecture. Design the number of CNN outputs based on the number of scenarios to be examined. Adopt the optimal hyperparameter combination for constructing CNN architecture.

Next, open the provided Jupyter Notebooks Step4_CNNmodel_ExternalValOriginal. ipynb and Step5_CNNmodel_ExternalValFlipped.ipynb. Validate the attained model using the independent images from the second replicate of the animal behavioral experiment.

Deploy the attained and validated model for classifying the video frame images generated from the experimental group using Jupyter Notebook Step6_CNNmodel_Application.ipynb. The MGS score significantly increased in mice given a mixture of wasabi and peanut butter compared to those given only peanut butter. The MGS score showed no significant difference between the control group and mice exposed to alcohol vapor.

The MGS score of mice provided with wasabi and peanut butter significantly decreased when exposed to alcohol vapor compared to those not exposed.