Bone Marrow and Tumor Radiomics at ¹⁸F-FDG PET/CT: Impact on Outcome Prediction in Non–Small Cell Lung Cancer

Abstract

luorine 18-fluorodeoxyglucose (FDG) PET imaging is traditionally used for clinical staging, and tumor uptake has been shown to help predict recurrence in many solid cancers across a diverse set of studies (1-3). Other studies (4-7) investigated the clinical use of quantifying nontumor regions at FDG PET/CT to help predict cancer outcome. For example, bone marrow hypermetabolism is associated with a higher risk of mortality in lung cancer (6-8), and organs of the reticuloendothelial system (eg, spleen and liver) are associated with a higher risk of recurrence in breast cancer (4). To our knowledge, the underlying biologic mechanisms for these associations are not currently well defined, but putatively studied areas identify a systemic response to cancer where glucose metabolism plays a role.Numerous studies (5,9-11) have shown that the development and maintenance of a systemic inflammatory response is associated with poorer outcomes in patients with cancer. Not surprisingly, blood correlates of this response, as measured by a complete blood cell count in the clinic, including platelets, white blood cells, and lymphocyte ratios, are also associated with cancer outcomes (12).

CT texture analysis predicts abdominal aortic aneurysm post-endovascular aortic aneurysm repair progression

“…As stated in the previous literature, a www.nature.com/scientificreports/ relative volume change larger than 2% after EVAR may require re-intervention, while a smaller than 2% change is considered relatively stable and requires no surgical intervention 24,25 ; additionally, the average variability is less than 2% for well-trained evaluators 26,27 . Texture analysis is a promising method in risk prediction and has been applied in various medical fields [28][29][30] , however, only a few studies have been carried out on the prediction value of AAA progression. Carl W. Kotze and colleagues applied CT heterogeneity analysis in fifty small aortic aneurysms without surgical intervention and identified the predictive role of texture analysis for aneurysms at risk of expansion 31 .…”

“…Texture analysis is a promising method in risk prediction and has been applied in various medical fields 28 – 30 , however, only a few studies have been carried out on the prediction value of AAA progression. Carl W. Kotze and colleagues applied CT heterogeneity analysis in fifty small aortic aneurysms without surgical intervention and identified the predictive role of texture analysis for aneurysms at risk of expansion 31 .…”

Artificial intelligence applications for oncological positron emission tomography imaging

“…Prognosis: lung cancer [62,64,72] nasopharyngeal carcinoma [60] gastric cancer [75] esophageal cancer [54] linear combination of Shannon information Genetic status: cerebral gliomas [38] This method could test and rank each feature with test score, determining the best features for training [38].…”

“…Prognosis: rectal cancer [76] lung cancer [61,62,66] AdaBoost, SVM and RF performed better than ANN both in accuracy and area under the ROC curve [3]. are appropriate for predicting histological classification, we need more multicenter studies with larger samples for the external validation in the future research.…”

Bone Marrow Radiomics of T1-Weighted Lumber Spinal MRI to Identify Diffuse Hematologic Marrow Diseases: Comparison With Human Readings

“…Radiomics is a novel field in medical imaging that aims to utilize large amount of quantitative features in order to advance decision support. Previously, 18 F-FDG PET/CT radiomics from bone marrows were used to differentiate various types of diseases [5][6][7], and radiomics from T1 and T2-weighted MRI were used to discriminate bone chondrosarcoma [8], metastatic diseases [9], and osteoporosis [10]. However, the MRI-based radiomics studies often discriminated diseases with low predictive performance [10] or with performance only comparable with experienced radiologists [8].…”