P2454 - A New Era in Colorectal Cancer Diagnostics: EfficientNetB0’s Impeccable Performance in Histopathological Diagnosis

Annie Ajang, BS¹, Ramya Elangovan, ², Kavin Elangovan, ², Jansi Sethuraj, BSN, RN, CCRN³, Tirth Patel, MBBS⁴, Elangovan Krishnan, MBBS, PhD, MS⁵
1St. George's University School of Medicine, Phoenix, AZ; 2AIM DOCTOR, Houston, TX; 3University of Texas Health Science Center, Houston, TX; 4GMERS Medical College and Hospital, Himmatnagar, Himmatnagar, Gujarat, India; 5AIM DOCTOR, Chennai, Tamil Nadu, India

Introduction: Colorectal cancer ranks among the most prevalent and lethal malignancies worldwide, with rising incidence and substantial diagnostic challenges due to histological complexity and interobserver variability. Artificial intelligence, particularly deep learning, is revolutionizing cancer diagnostics by automating feature extraction and classification from complex histopathological images. EfficientNetB0, a state-of-the-art convolutional neural network, employs compound scaling to optimize model depth, width, and resolution, achieving superior accuracy and computational efficiency with fewer parameters than traditional architectures. Its application promises to elevate the precision and reproducibility of colorectal cancer detection.

Methods: A large, anonymized dataset of 10,000 histopathological colon tissue images, encompassing malignant and benign cases, was randomly assigned into training (60%), validation (20%), and testing (20%) cohorts. Images underwent comprehensive preprocessing and augmentation to maximize model robustness. The EfficientNetB0 architecture was trained and optimized, with diagnostic performance assessed using accuracy, precision, recall, F1-score, F2-score, and area under the receiver operating characteristic curve (AUROC) on validation and test sets. The final model was deployed in a universally accessible, cross-platform application for real-time expert validation and diagnostic support.

Results:

EfficientNetB0 achieved exemplary diagnostic performance in classifying colorectal tissue. The model attained an exceptional accuracy, precision, and recall greater than 99% for both malignant and benign cases. The F1-score was 1.000 for both cancer and normal tissue. The AUC reached 1.000, underscoring the model’s discriminative power. These results highlight EfficientNetB0’s exceptional capacity to deliver robust, reproducible diagnostics in histopathology.

Discussion: EfficientNetB0’s high accuracy and efficiency exemplify a transformative leap in colorectal cancer diagnostics. By automating nuanced feature extraction, EfficientNetB0 reduces diagnostic subjectivity, accelerates clinical workflows, and enables equitable access to advanced diagnostics. For clinicians, integrating such AI tools augments decision-making and may improve outcomes through earlier, more precise detection of colorectal malignancy.




Disclosures:


Annie Ajang, BS¹, Ramya Elangovan, ², Kavin Elangovan, ², Jansi Sethuraj, BSN, RN, CCRN³, Tirth Patel, MBBS⁴, Elangovan Krishnan, MBBS, PhD, MS⁵. P2454 - A New Era in Colorectal Cancer Diagnostics: EfficientNetB0’s Impeccable Performance in Histopathological Diagnosis, ACG 2025 Annual Scientific Meeting Abstracts. Phoenix, AZ: American College of Gastroenterology.