Estimating the size of polyps during actual endoscopy procedures using a spatio-temporal characterization

Area of Science:

• Medical Imaging
• Gastroenterology
• Computational Pathology

Background:

• Colorectal polyps, particularly those >10mm, require accurate size estimation for determining post-polypectomy surveillance intervals.
• Current polyp size measurement during colonoscopy is challenging due to factors like anatomical variability, illumination changes, and camera movement.
• Precise polyp size measurement is critical for patient management and preventing colorectal cancer progression.

Purpose of the Study:

• To develop and validate a semi-automatic method for accurate colorectal polyp size estimation using endoscopic video sequences.
• To assess the performance of the proposed method against expert measurements and simulated polyp models.
• To improve the reliability of polyp size measurements for guiding patient follow-up schedules.

Main Methods:

• A semi-automatic method was developed to estimate polyp size by propagating an initial manual delineation across video frames using spatio-temporal characteristics.
• The method was evaluated on real endoscopy video sequences and simulated polyp models with known sizes.
• Performance was quantified using Dice Score and Root Mean Square Error (RMSE), and compared against measurements from multiple expert endoscopists.

Main Results:

• The method achieved a Dice Score of 0.7 when compared to expert delineations in real endoscopy videos.
• In simulated environments, the method yielded an RMSE of 0.87mm for size estimation.
• Comparison with four expert endoscopists on real sequences (polyp sizes 5-20mm) resulted in an RMSE of 4.7mm, with no statistically significant difference (p<0.01) from expert measurements.

Conclusions:

• The proposed semi-automatic method provides accurate and reliable colorectal polyp size estimations during routine endoscopic examinations.
• The method's performance is comparable to experienced endoscopists, offering a potential tool to standardize polyp size assessment.
• This technique can aid in optimizing surveillance intervals, thereby improving patient management and colorectal cancer prevention strategies.