AI-Powered Predictions of Breast Ductal in Situ Carcinoma Morphology and Surgical Outcomes.
Artificial intelligence model predicted stromal invasion in ductal carcinoma in situ with 72.4% accuracy
AI-Powered Predictions of Breast Ductal in Situ Carcinoma Morphology and Surgical Outcomes.
Artificial intelligence (artificial intelligence) is increasingly integrated into oncological imaging, but its ability to predict detailed histopathological features from standard mammography remains understudied in ductal carcinoma in situ (ductal carcinoma in situ).
This study aimed to evaluate the performance of a large language model (ChatGPT-4, Open artificial intelligence, May 2025) in predicting nuclear grade, architectural subtype, comedo necrosis, and stromal invasion from specimen mammography.
We conducted a retrospective and methodological study of 29 patients with histologically confirmed ductal carcinoma in situ or invasive carcinoma with ductal carcinoma in situ components.
correctly ruled out invasion far more often than it caught true invasion
The size of mammographic lesions ranged from 1.2 to 10.0 mm (mean Ã+- SD: 4.46 Ã+- 2.25 mm).
Histopathological diagnoses included pure ductal carcinoma in situ (n = 17), invasive NST carcinoma with ductal carcinoma in situ (n = 10), and mixed histologies (n = 2).
Nuclear grade classification matched histopathology in 20.7% of cases, while architectural subtype classification achieved 17.2% agreement.
Multiclass predictions showed low F1 scores for most categories.
Although the artificial intelligence model demonstrated acceptable utility for detection of comedo necrosis and excluding stromal invasion, it faced several difficulties regarding nuclear grading and architectural subtype classification.
Although limited by the small sample size and 2D imaging, this methodological study provides an insight for future artificial intelligence and radiomics approaches in breast tumor characterization.