Lung Cancer Diagnosis using Artificial Intelligence
Principal Investigator
Name
Shital Bhatt
Degrees
M.E.
Institution
MBIT
Position Title
ASSISTANT PROFESSOR
Email
About this CDAS Project
Study
NLST
(Learn more about this study)
Project ID
NLST-547
Initial CDAS Request Approval
Jul 29, 2019
Title
Lung Cancer Diagnosis using Artificial Intelligence
Summary
I, Shital Bhatt currently pursuing Ph.D. from Gujarat Technological University, Gujarat, India. My topic of research is "Lung Cancer Diagnosis using Artificial Intelligence". I would be using CT scan data of patients who have nodules less than 20mm. In this research, I am aiming to detect the possibility of early-stage cancer in the lungs. I would be building a Hybrid 3D CNN Deep Learning model whose accuracy, precision and sensitivity can be improved by using Machine Learning along with Deep Learning.
The model will be the ensemble of many different 3D CNN along with Machine Learning Algorithm. I would like to diagnose the benign or malignant possibility and further to it the type of lung cancer and its exact location in lungs so as to help the radiologists and oncologist to further use it to attain biopsy sample. I am planning to advance the model to Transfer Learning concept where the model designed for the diagnosis of Lung Cancer can be used for another similar category of cancer. The model can be converted to a Software designed on AI builder with a backend of Tensor Flow which can be connected to the Medical Imaging software where CT Scan images are captured so as to make it more user-friendly. This concept can be further taken to Reinforcement Learning in order to make it a Real-Time Application.
My aim is to build a Model and deploy it on software which can be used by radiologist and oncologist at rural areas, so that the patient need not to travel long journey to get the diagnosis.
Firstly I want to do diagnosis on CT scan images and Pathology images separately and than to improve the performance of model I would like to do fusion of both models.
Diagnosis from Pathology images will be quite useful as at present as individually the Pulmonary nodules are not accurately defined on CT images and so lead to unnecessary false positive ratio.
The World Statistics as shown by WHO in the year 2019 mentions that the most deadly cancer among 100 different types is Lung Cancer and which is too difficult to be detected at an early stage.
The model will be the ensemble of many different 3D CNN along with Machine Learning Algorithm. I would like to diagnose the benign or malignant possibility and further to it the type of lung cancer and its exact location in lungs so as to help the radiologists and oncologist to further use it to attain biopsy sample. I am planning to advance the model to Transfer Learning concept where the model designed for the diagnosis of Lung Cancer can be used for another similar category of cancer. The model can be converted to a Software designed on AI builder with a backend of Tensor Flow which can be connected to the Medical Imaging software where CT Scan images are captured so as to make it more user-friendly. This concept can be further taken to Reinforcement Learning in order to make it a Real-Time Application.
My aim is to build a Model and deploy it on software which can be used by radiologist and oncologist at rural areas, so that the patient need not to travel long journey to get the diagnosis.
Firstly I want to do diagnosis on CT scan images and Pathology images separately and than to improve the performance of model I would like to do fusion of both models.
Diagnosis from Pathology images will be quite useful as at present as individually the Pulmonary nodules are not accurately defined on CT images and so lead to unnecessary false positive ratio.
The World Statistics as shown by WHO in the year 2019 mentions that the most deadly cancer among 100 different types is Lung Cancer and which is too difficult to be detected at an early stage.
Aims
Detecting the Early Lung Cancer Nodule.
Diagnosing whether the Nodule is cancerous or non-cancerous.
Finding the type of Lung Cancer.
Find and show 3D view of Nodule which can be helpful to radiologist and oncologist during the biopsy.
The model can be built-in AI Builder whose backend programming will be on TensorFlow.
Using the Model for Real-Time application.
Collaborators
DR. HIMANSHUKUMAR BIPINBHAI SONI, PRINCIPAL, 011 - G H PATEL COLLEGE OF ENGINEERING & TECHNOLOGY Dr. Heena Rahul Kher, Assistant Professor, A D Patel Institute of Technology (ADIT)
Dr. Tanmay Pawar, Professor, B V M Engineering College