Adopting transfer learning for neuroimaging: a comparative analysis with a custom 3D convolution neural network model.
Amira SolimanJose R ChangKobra EtminaniStefan ByttnerAnette DavidssonBegoña Martínez-SanchisValle CamachoMatteo BaucknehtRoxana StegeranMarcus RessnerMarc Agudelo-CifuentesAndrea ChincariniMatthias BrendelAxel RomingerRose BruffaertsRik VandenbergheMilica G KrambergerMaja TrostNicolas NicastroGiovanni B FrisoniAfina W LemstraBart N M van BerckelAndrea PilottoAlessandro PadovaniSilvia MorbelliDag AarslandFlavio NobiliValentina Garibottonull nullMiguel Ochoa-FigueroaPublished in: BMC medical informatics and decision making (2022)
TL can indeed lead to superior performance on binary classification in timely and data efficient manner, yet for detecting more than a single disorder, TL models do not perform well. Additionally, custom 3D model performs comparably to TL models for binary classification, and interestingly perform better for diagnosis of multiple disorders. The results confirm the superiority of the custom 3D-CNN in providing better explainable model compared to TL adopted ones.