Author: nkamona

Our PhD student, Shuyue (Frank) Guan, has attended the 47th Annual IEEE AIPR 2018: Ubiquitous Imaging in Washington DC, October 9th – 11th. Frank gave a presentation about segmentation of thermal breast images using Convolutional and Deconvolutional Neural Networks.

The Applied Imagery Pattern Recognition (AIPR) workshop sponsored by IEEE is to bring together researchers from government, industry, and academia across a broad range of disciplines. The 2018 IEEE AIPR Workshop explored reemerging intersections and synergies between imaging, Big Data and cloud compute & hardware advances, continuing the workshop’s long tradition of bringing together researchers and developers who span the disciplines and work in labs across academia, industry, and government.

Here is a brief summary of Frank’s project and presentation:

We are investigating infrared thermography as a noninvasive adjunctive to mammography for breast screening. Thermal imaging is safe, radiation-free, pain-free, and non-contact. Segmentation of breast area from the acquired thermal images will help limit the area for tumor search and reduce the time and effort needed for manual hand segmentation. Autoencoder-like convolutional and deconvolutional neural networks (C-DCNN) are promising computational approaches to automatically segment breast areas in thermal images. In this study, we apply the C-DCNN to segment breast areas from our thermal breast images database, which we are collecting in our clinical trials by imaging breast cancer patients with our infrared camera (N2 Imager). For training the C-DCNN, the inputs are 132 gray-value thermal images and the corresponding manually-cropped breast area images (binary masks to designate the breast areas). For testing, we input thermal images to the trained C-DCNN and the output after post-processing are the binary breast-area images. Cross-validation and comparison with the ground-truth images show that the C-DCNN is a promising method to segment breast areas. The results demonstrate the capability of C-DCNN to learn essential features of breast regions and delineate them in thermal images.

Check the poster for more information of this project.

Congratulations to Dr. Murray Loew! He and his collaborator, Dr. Sharad Goyal (Radiation Oncology), have received second-year, competitive funding from The George Washington University Cross-Disciplinary Research fund for their project, “Development of a novel radiomics platform to predict outcomes in advanced head and neck cancer.”

GW Cancer Center Basic Sciences Retreat featured talks and posters by trainees (graduate students, postdocs, staff scientists, technicians, lab managers, and other full time staff) that took place on Friday, April 27, 2018 at GW Cancer Center. Our Master’s student, Apurva Singh, presented her project “Effectiveness of Radiation Therapy for the Treatment of Head and Neck Squamous Cell Carcinoma“.

The IEEE International Symposium on Biomedical Imaging (ISBI) is a scientific conference dedicated to mathematical, algorithmic, and computational aspects of biological and biomedical imaging, across all scales of observation. Our graduate students Shuyue Guan and Kristina Landino have presented their posters in the conference that took place on April 4-7th in Washington, DC.

Shuyue presented his work on mammographic detection based on the computer-aided diagnosis (CAD). Below is a summary of his work:

“Mammographic detection based on the computer-aided diagnosis (CAD) can improve treatment outcomes for breast cancer and longer survival times for the patients. For the breast cancer detection, the Convolutional Neural Network (CNN) can extract features from mammographic images automatically and then do classification. However, to train the CNN from scratch needs a huge number of labeled images, which is infeasible for some kinds of medical image data such as mammographic tumor images. A promising solution is to apply transfer learning in CNN. In this study, we applied the pre-trained VGG-16 model to extract features from input mammographic images and used these features to train a Neural Network (NN)-classifier. On the DDSM database, average validation accuracy converged at about 0.905 for abnormal vs. normal cases with 10-fold cross validation, and no obvious overfitting happened. Therefore, this study shows that applying transfer learning in CNN can detect female breast cancer from mammogram, and training a NN-classifier by feature extraction is a feasible method in transfer learning.”

Check the posters for more information here.