The most prevalent metastatic location for rectal carcinoma (RC) are lymph nodes (LNs), and the nodal status is crucial to treating and forecasting choices. The site and several metastatic LNs should be investigated before treatment guidelines comply with the NEC Network and the American Joint Committee on Cancer (AJCC) stage standards. Identifying and removing metastatic LNs during the intervention is crucial to prevent tumour repetition, especially in lateral lines. Some studies have shown that stronger LLNs can be closer to local recurrence and showed that dissection of LLNs might enhance prognosis and reduce local recurrence for patients with poor RC at these locations. In contrast, LLND is an autonomous procedure with generally more surgical implications, including surgery and long-term sexual and urinary problems. Therefore, the correct number and position of metastatic LNs must be indicated before surgery in the therapy option. This work deals with segmentation and classification challenges for normal and abnormal lymph nodes. Here Preprocessing is performed by Curvature based shearlet filter with Contrast Limited Savitzky-Golay Histogram Equalization is used. The Semi-Supervised Fuzzy Logic clustering Algorithm was then used to segment lymph nodes. Once the lymph region is segmented, the grey level co-occurrence matrix is used to extract functions (GLCM). Then the scale-down bee herd optimization approach is introduced to minimize the number of measures by features, which increases the classifier detection rate. Deep residual Boltzmann Convolution Neural Network system of classification creates a pattern for the benign lymph nodes and the malignant identification. In comparison to state-of-the-art procedures, the simulation results demonstrate the importance of the proposed method by obtaining a high range of precision (97.7%), sensitivity (95.7%) and specificity (95.8%) than any other existing methodology

Rectal cancer, Lymph nodes, Curvature based shearlet filter, Contrast Limited Savitzky-Golay Histogram Equalization, Semi-Supervised Fuzzy logic clustering algorithm, Gray Level Co-occurrence Matrix, Scale down Bee herd colony optimization, Deep residual Boltzmann Convolution Neural Network.