Evaluating the Impact of a Ki-67 Decision Support Algorithm on Pathology Residents' Scoring Accuracy

Area of Science:

• Oncology
• Pathology
• Medical Imaging

Background:

• Accurate Ki-67 scoring is crucial for breast cancer evaluation and treatment decisions.
• Pathology residents often face challenges in achieving consistent Ki-67 assessments.

Purpose of the Study:

• To evaluate a Ki-67 algorithm as a decision support tool for pathology residents.
• To improve the accuracy and consistency of Ki-67 scoring in breast cancer diagnosis.

Main Methods:

• Retrospective analysis of Ki-67 scores on whole slide images (WSI) from 156 breast cancer patients.
• Comparison of senior pathologist scores, algorithm-generated scores, and resident scores (with and without algorithm assistance).
• Statistical analysis using Wilcoxon and intra-class correlation (ICC) tests.

Main Results:

• The Ki-67 algorithm showed high correlation with senior pathologist scores (ICC, 0.98).
• Residents using the algorithm achieved scores significantly closer to senior pathologists' evaluations (p=0.008).
• Residents modified their scores in 53.8% of cases, with 74% of improved scores showing an increase.

Conclusions:

• The Ki-67 algorithm effectively serves as a decision support tool for pathology residents.
• Algorithm assistance enhances resident accuracy and alignment with expert pathologist assessments in Ki-67 scoring.
• This tool has the potential to improve diagnostic consistency in breast cancer pathology.