Table 2.

Deep learning systems in therapeutic applications

Network Types	Key Characteristics	Detailed Description	Notable Remarks
Deep auto-encoder [25]	The input–output layers of the framework have a break indeed with the number of neurons, which must be at least 2	Utilized for unsupervised learning, connected for dimensionality decrease or change	Utilized for include extraction and determination
Deep Boltzmann Machine [26]	The layers in the address are undirected and have a measurement of 2 or more. These layers can be categorized as obvious or covered up, with no nearness of input or yield layers	The undirected associations encourage both administered and unsupervised learning, whereas too minimizing the time required for the learning handle	Not appropriate for huge datasets
Convolutional Neural Networks [27]	Incorporates classification, convolution, pooling, and completely connected layers. Utilize an enactment work that's not straight. Acknowledges input specifically as an image	Utilized to fathom therapeutic image categorization issues for unremitting illness and cancer discovery	Apply the network's highlight extraction preparation. Not each neuron is wired together. takes too much data to memorize
ResNet-50 [28]	A 50-layer deep CNN sort that's more advanced comprises leftover units that skip associations	Utilized to classify therapeutic images with moved forward execution Requires more preparing time than CNN but performs impressively superior	It takes much data as well to memorize
GoogLeNet [29]	Inception CNN: concurrent convolution with diverse part sizes, high-performance therapeutic image classification	In Initiation, CNN trains sped up more than ResNet-50, even though its execution was a little better	Request a gigantic set of data to memorize effectively
EfficientNet [30]	CNNs can make strides by expanding their profundity, width, and determination	Utilized to fathom a part of the image categorization issue. Compared to ResNet-50 and 101	It is smaller and speedier