Posted: Wednesday, September 13, 2023
In a study published in the journal Military Medical Research, Chinese researchers explored whether artificial intelligence (AI) methodology based on MRI can aid in the management of prostate cancer. Researchers also explored the question of whether AI procedures are generally superior to common clinical assessment methods for diagnosing prostate cancer.
“This study focused on two fields of AI application to [prostate cancer]: the diagnosis field, which refers to the identification of malignant lesions and stratification of [prostate cancer] risk; and the prediction field, which refers to the prediction of therapeutic outcomes or progression of [prostate cancer],” wrote study authors Liu et al.
Researchers summarized multiple studies comparing the diagnostic and predictive performance for prostate cancer between AI and common clinical assessment methods based on MRI and clinical characteristics. They discovered that AI methods achieved a higher summary receiver operator characteristic curve than clinical assessment methods for the diagnosis of prostate cancer. AI methods also achieved a higher summary receiver operating characteristic area under the curve than clinical assessment methods when predicting adverse pathology. The study presented a relatively high total average radiomics quality score of 15.2.
The scores of the individual radiomics quality score elements suggest the AI models in these studies were constructed with relatively perfect and standard radiomics processes. Researchers further postulate that the exact generalizability and clinical practicality of the AI models should be further validated using higher levels of evidence, such as prospective studies and open-testing data sets.
Disclosure: The study authors reported no conflicts of interest.