A convolutional neural network based approach for generating full PET/CT image series from shorter scan time

2023 Fiscal Year Research-status Report

• Project/Area Number: 19K20685
• Research Institution: Chiba University
• Principal Investigator: BhusalChhatkuli Ritu 千葉大学, 子どものこころの発達教育研究センター, 特任助教 (50836591)
• Project Period (FY): 2019-04-01 – 2025-03-31
• Keywords: PET/CT / CNN-LSTM / Neural Network / Prediction

Outline of Annual Research Achievements

Initial and delayed scans, also known as dual-time-point scans, are widely used in positron emission tomography/computed tomography (PET/CT) for the diagnosis and delineation of pancreatic cancer; however, their acquisition is relatively time-consuming. The purpose of this pilot study was to use neural network based method to eliminate the need of delayed PET/CT scan for diagnosis. The results obtained from our CNN-LSTM based analysis suggested that our study could obviate the need for delayed scans however validations studies are required for the clinical application of our model.

Current Status of Research Progress

4: Progress in research has been delayed.

Reason

The pilot study (Data acquisition and analysis) has been completed and the journal paper was submitted, unfortunately it could not be accepted hence we are currently correcting the manuscript and preparing for submission in other journal.

Strategy for Future Research Activity

Currently the manuscript is being prepared for this work and is to be submitted in a scientific journal.

Causes of Carryover

We are yet to publish our paper in a scientific journal and also planning to present the latest result in a conference hence the expenses are needed to be carried to next fiscal year.