An integrated 3D-sparse deep belief network with enriched seagull optimization algorithm for liver segmentation

Customized m-RCNN and hybrid deep classifier for liver cancer segmentation and classification

“…Group 1 utilized CNN for ten samples, and Group 2 used SVM with 0.8 G power for ten samples. The precise elliptical liver structure is extracted using a 3D Sparse Deep Belief Network with Enriched Seagull Optimization (3D-SDBN-ESO) that was developed by Dickson et al [ 25 ]. The preparation stage of the liver segmentation procedure begins with the feeding of the abdominal CT images.…”

“… These systems were deficient in segmenting and detecting liver lesions due to low contrast between the liver and the neighboring organs. 2023 Jun 22 Dickson et al [ 25 ] 3D-SDBN Accurately segments the liver from abdominal CT images. Vascular structures must be addressed, and complete segmentation with lesions must be achieved.…”

“…Additionally, we evaluated the effectiveness of adaptive histogram normalization using metrics such as PSNR and SSIM. Furthermore, we compared the performance of the SqueezeNet + DeepMaxout method with state-of-the-art approaches like m -RCNN [ 10 ], Hybrid ResUNeT [ 12 ], SVM [ 24 ], and 3D-SDBN [ 25 ], as well as traditional methods such as KNN, LSTM, GRU, DeepMaxout, SqueezeNet, and DCNN. Fig.…”

“…Fig. 9 (a–c) presents a comparative analysis of positive metrics for liver cancer classification between the SqueezeNet + DeepMaxout method and a set of other models, including KNN, LSTM, GRU, DeepMaxout, SqueezeNet, DCNN, m -RCNN [ 10 ], Hybrid ResUNeT [ 12 ], SVM [ 24 ], and 3D-SDBN [ 25 ]. Upon a thorough examination of all the positive metric graphs, it becomes evident that the SqueezeNet + DeepMaxout approach outperforms the others, achieving superior values and providing precise classifications for liver cancer.…”

“…10 . Furthermore, the SqueezeNet + DeepMaxout approach is compared to KNN, LSTM, GRU, DeepMaxout, SqueezeNet, DCNN, m -RCNN [ 10 ], Hybrid ResUNeT [ 12 ], SVM [ 24 ], and 3D-SDBN [ 25 ]. Minimizing the rate of negative metrics is crucial for efficient liver cancer classification.…”