Amal Ibrahim

Sudan

Application of Artificial Intelligence in the Diagnosis and classification ovarian cancer in Sudanese

Amal Jamal Balla Ibrahim 1,Rawng Al-zain Adam Hussein 1,Anas Salih Suliman Abdalla 1
Department of Medical Laboratory science (Histopathology and Cytology),Al Imam Alhade Golleg

Abstract

Background

Ovarian cancer remains one of the deadliest gynecologic malignancies worldwide, with a particularly high mortality rate in Sudan due to late diagnosis and limited diagnostic resources. It originates in the ovaries and can spread to the fallopian tubes and peritoneum, often remaining asymptomatic until advanced stages.

Methods

Methods:
A descriptive cross-sectional study using the Dataset Collection as Histopathological pictures of ovarian tissue were obtained from the oncology departments of Kosti, Gedaref, and Kassala hospitals. The dataset includes both benign and malignant cases, which were annotated by medical personnel. Pre-processing: Images were scaled.

Results

Two robust ML algorithms—Support Vector Machine (SVM) and K-Nearest Neighbor (KNN)—were initially applied for image-based classification, with SVM achieving superior performance (98.1% training accuracy and 97.16% validation accuracy). Meanwhile, a Random Survival Forest (RSF) model was used to assess the influence of various clinical features on survival. SHAP (Shapley Additive Explanations) was utilized for model interpretability, pinpointing tumor residual status, patient age, and biomarker levels (CA125 and HE4) as the most impactful factors. A web-based tool was also created to enable real-time classification using image input and biomarker data.

Conclusions

Conclusion:
This integrated, explainable AI-based approach shows great promise for improving early detection, guiding treatment choices, and boosting survival outcomes in the resource-limited Sudanese healthcare sector in Sudan.