A Comprehensive Study on Deep Learning Models for the Detection of Ovarian Cancer and Glomerular Kidney Disease using Histopathological Images

Ovarian cancer is a significant health concern because of its high mortality rates and potential to cause glomerular injury, which can obstruct the urinary tract. It is very crucial to diagnose and treat these diseases accurately as well as timely. In the era of artificial intelligence, deep learning models have emerged as powerful tools in analysing medical images as they showcase exceptional capabilities to detect diseases. In this study, an innovative approach has been proposed that uses deep transfer learning classifiers for the detection as well as classification of ovarian cancer, sclerosed glomeruli, and normal glomeruli in histopathological images. To gather relevant data, two different repositories have been explored which contain images of ovarian cancer, sclerosed glomeruli, and normal glomeruli. These images are thoroughly pre-processed by converting them into grayscale. Afterwards, advanced segmentation techniques are applied such as image equalization, thresholding, image inversion, and morphological opening which effectively highlight the affected areas using contour features, and various measurements such as area, mean intensity, height, width, and epsilon are calculated. Our study employed a range of deep learning techniques such as AlexNet2, InceptionV3, EfficientNetB0, EfficientNetB5, DenseNet121, Xception, MobileNetV2, and InceptionResNetV2 along with the two optimization techniques: Adam and RMSprop optimizer. Remarkably, during experimentation, AlexNet2 demonstrated exceptional accuracy by achieving 99.74%, with a low loss of 0.0018 and a root mean square error of 0.042426 when incorporating the Adam optimizer. Similarly, using the RMSprop optimizer, Xception delivered outstanding results with an accuracy of 99.74%, a minimal loss of 0.0027, and a root mean square error of 0.051962. This pioneering research significantly contributes to the field of medical diagnostics by harnessing deep learning technology to enhance the precision and efficiency of ovarian cancer and sclerosed glomeruli detection.