- Developed a novel DCNN-based approach for common thoracic disease diagnosis using CXRs.- Achieved state-of-the-art results on the Chexpert benchmark dataset for the classification task.

- Built a high-precision NLP tool for automatic extraction of negative and uncertain findings in radiology reports.- Achieved significant improvements over prior benchmarks on all datasets and the tool is currently being used for generating labels for training DCNN models for abnormality detection.

- Developed and implemented a novel neural network approach for accurate and efficient detection of Persistent Scatterers (PS) in PSInSAR processing.- Designed a CNN-based denoising autoencoder to improve the quality of interferograms by effectively filtering out noise.- Utilized an LSTM-based network to model the temporal behavior of stochastic noise in PSInSAR data, leading to more accurate coherence prediction.