Sahar Mansour, Professor of Radiology at Kasr Al-Ainy Hospital, Cairo University, shared on LinkedIn:
“AI is everywhere.
But is it for your patient population?
For years, healthcare systems across Middle East and Africa have adopted AI tools trained predominantly on European, Asian, or North American datasets.
The challenge is not the technology.
The challenge is that diseases, genetics, breast density patterns, lifestyle, and epidemiology differ across populations.
An AI system developed through collaboration between Applied Innovation Center – AIC and Baheya Foundation demonstrated the value of training AI on regional datasets rather than relying solely on imported models.
In heterogeneously dense breast, a common challenge in our region, the system successfully identified left breast abnormality with confidence 92% later confirmed as invasive lobular carcinoma (figure1), one of the most difficult breast cancers to detect.
Equally important, the AI did not overcall findings in the contralateral breast despite contrast enhanced mammogram abnormal contrastuptake (circle in figure2) that raised concern.
Surgical pathology subsequently confirmed benign proliferative changes, of fibrocystic disease and sclerosing adenosis.
Following treatment, left breast muscle flap reconstruction and right local excision follow-up, the system continued to appropriately classify expected postoperative changes without confusing scar tissue for recurrent disease helping reduce unnecessary alarms and false-positive interpretations (arrows in figure3).
This is what excites me about the future: AI that enhances workflow, reduces cognitive bias and fatigue while improving precision and confidence.
Speaking healthcare is not about having more AI.
It is about precision medicine which begins with population-specific intelligence.
Special acknowledgment to Dr. Ahmed Rozeka, Dr.Amr Moustafa, and Dr. Mohammed Gomaa for their expert supervision of the AI training process that made this work possible.”
Other articles about Breast Cancer on OncoDaily.
