Date of Award

August 2022

Degree Type


Degree Name

Doctor of Philosophy



First Advisor

Zeyun Yu

Committee Members

Sandeep Gopalakrishnan, Rohit Kate, Jun Zhang, Tian Zhao


Deep Learning, Medical image classification, Multi-modality, Osteosarcoma classification, Wound classification, Wound localization


Medical image classification is an essential part of diagnosis, which with automation may benefit both physicians and patients in terms of time and cost. For automation, different Artificial intelligence (AI) methods, including Machine Learning (ML) and Deep Learning (DL), are used widely. Specifically, DL algorithms have become popular in classifying medical images due to their propensity for good performance. This thesis studies medical image classification problems using deep learning models. Four specific medical applications are considered: (1) Osteosarcoma cancer classification in histological images, (2) Burn wound classification, (3) Wound severity classification from clinical images, and (4) Wound type classification using wound images and their corresponding locations. Alongside these classifications, a pre-processing task of automatic wound region of interest (ROI) detection is also performed using a deep neural network. Transfer learning models are used for osteosarcoma classification due to the scarcity of data, and state-of-the-art performance is achieved. In burn wound and wound severity classifications, transfer learning, end-to-end learning, and stacked deep learning models are used. Both classifications show promising performance. Finally, a novel deep learning multi-modality model is developed to handle image and categorical modalities. This model takes wound images and their corresponding locations as input and predicts the wound types based on the information from both modalities. State-of-the-art performance is achieved with this developed network. Public datasets are used for osteosarcoma and burn wound classifications. Wound datasets are designed for wound localization, wound severity, and wound type classifications with the supervision of wound specialists. A body map is also developed for wound location labeling.

Available for download on Tuesday, February 28, 2023