Improving Colorectal Cancer Diagnosis Using MIRNet and InceptionV3 on Histopathological Images

• Neilson P. Ribeiro IFMA / UFMA
• Felipe R. S. Teles UFMA
• João Otávio Beira Diniz IFMA
• Luana B. da Cruz UFCA
• Domingos A. Dias Jr. UFCA
• Geraldo Braz Junior UFMA
• João D. S. de Almeida UFMA
• Anselmo C. de Paiva UFMA

Resumo

Colorectal cancer (CRC) is the second most prevalent type of cancer among men and women in Brazil, accounting for 9.1% of cancers in men and 9.2% in women between 2020 and 2022. CRC diagnosis typically relies on histopathological images, which can be challenging to interpret and prone to errors. This paper proposes a method for diagnosing CRC using histopathological images. This method involves enhancing images with MIRNet and classifying them as benign or malignant using InceptionV3. The results surpass the state-of-the-art by introducing, for the first time, enhancement through MIRNet, achieving an accuracy of 99.67% and an AUC of 0.9999. We believe that this method, combined with medical expertise, holds great promise for early CRC diagnosis.