Researchers are evaluating how machine learning can be applied to the investigation of medical imaging biomarkers, such as CT scans, to detect and diagnose lung diseases.
Medical researchers specializing in radiology published a review in The Lancet Digital Health where they seek to evaluate the use of tools of machine learning for the early detection of interstitial lung diseases or ILD.
The evaluation of interstitial lung disease (ILD) focuses on diagnoses dependent on clinical, radiological, and pathological processes. However, according to specialists, these do not reliably describe the behavior of the diseases. The evaluation of these diseases requires integrating clinical and imaging data, and in certain cases, even biological material such as surgical lung biopsy, cryobiopsy, and others.
Therefore, the researchers conducted this review on the use of tools Artificial Intelligence (AI), as are the machine learning and the deep learning. The algorithms of machine learning They can identify ILD in at-risk populations, predict the extent of pulmonary fibrosis, correlate radiological abnormalities with the deterioration of lung function, and even be used as endpoints in treatment trials.
The use of machine learning It is based primarily on image analysis. Medical image datasets are very extensive, and clinically relevant information is unstructured. This is where the [technique/technique] comes into play. machine learning, to organize unlabeled data and extract features to facilitate analysis.
“The implementation of machine learning and the deep learning "It has greatly improved the efficiency, accuracy, and reproducibility of various segmentation methods," the authors explain.
Feature extraction by deep learning This refers to mathematical operations performed on digital images to produce numerical descriptors of texture, shape, and other distinctive characteristics. These are then computationally analyzed using clinical parameters.
Learn more by reading the full review:
https://www.thelancet.com/journals/landig/article/PIIS2589-7500(22)00230-8/fulltext
