Bolin Song
BME PhD Proposal Presentation

Date: 2023-06-08
Time: 11:00 - 11:30 am
Location / Meeting Link: https://teams.microsoft.com/l/meetup-join/19%3ameeting_MWJhNWMyOTAtOWYzNS00NTZhLTgxZTMtNzAyMjRiNDhhMTRj%40thread.v2/0?context=%7b%22Tid%22%3a%22e004fb9c-b0a4-424f-bcd0-322606d5df38%22%2c%22Oid%22%3a%22a8bfc537-8e8f-47f4-a98d-af15c10e49d4%22%7d

Committee Members:
Prof. Ahmet F. Coskun; Prof. May Dongmei Wang; Prof. Saurabh Sinha; Prof. Nabil F. Saba; Prof. Anant Madabhushi;


Title: Computational imaging-based prognostic and predictive biomarker to predict survival outcome and added benefit of chemotherapy in oropharyngeal cancer

Abstract:
It is estimated that 54,540 adults in the United States will be diagnosed with oropharyngeal squamous cell carcinoma (OPSCC) in 2023, from which 70% [1] will be Human papillomavirus related (HPV+). HPV+ OPSCC differs from those tobacco or alcohol related (HPV-) OPSCC in disease aggressiveness and response to treatment [2], leading to different treatment paradigms. Thus, separate risk stratification strategies should be provided to these two different cancer entities for guiding treatment planning. Regarding the treatment options, radiotherapy or concurrent chemoradiotherapy are two common treatment strategies for early stage (I, II) HPV+ OPSCC patients. However, some patients will not receive additional benefits from the chemotherapy after radiation [3]. Therefore, it is critical to identify those HPV+ patients in early stage who will likely not benefit from the additional chemotherapy and are suitable for treatment de-escalation. Radiologic images such as CT and pathology Whole Slide Images (WSI) are intensively studied for outcome and treatment response predictions, providing enriched prognostic information on anatomical (organ) and morphological (cellular) levels, respectively. However, most of the existing studies only focus on primary tumor from a single image modality [4,5,6,7], without utilizing the complementary information across image modalities (i.e. CT and WSI) and regions of interest (i.e. primary tumor and lymph nodes). To this end, this work aims to develop and validate three computational workflows: • A prognostic tool for OPSCC risk-stratifications taking HPV status into consideration • A predictive tool to identify potential early-stage OPSCC candidates who will not benefit from chemotherapy • A multimodal machine learning framework to effectively combine the information from radiology and pathology imaging data for outcome predictions