Background: Radiation pneumonitis (RP) is a dose-limiting toxicity in lung cancer radiotherapy (RT). As risk factors in the development of RP, patient and tumor characteristics, dosimetric parameters, and treatment features are intertwined, and it is not always possible to associate RP with a single parameter. This study aimed to determine the algorithm that most accurately predicted RP development with machine learning. Methods: Of the 197 cases diagnosed with stage III lung cancer and underwent RT and chemotherapy between 2014 and 2020, 193 were evaluated. The CTCAE 5.0 grading system was used for the RP evaluation. Synthetic minority oversampling technique was used to create a balanced data set. Logistic regression, artificial neural networks, eXtreme Gradient Boosting (XGB), Support Vector Machines, Random Forest, Gaussian Naive Bayes and Light Gradient Boosting Machine algorithms were used. After the correlation analysis, a permutation-based method was utilized for as a variable selection. Results: RP was seen in 51 of the 193 cases. Parameters affecting RP were determined as, total(t)V5, ipsilateral lung D_max, contralateral lung D_max, total lung D_max, gross tumor volume, number of chemotherapy cycles before RT, tumor size, lymph node localization and asbestos exposure. LGBM was found to be the algorithm that best predicted RP at 85% accuracy (confidence interval: 0.73-0.96), 97% sensitivity, and 50% specificity. Conclusion: When the clinical and dosimetric parameters were evaluated together, the LGBM algorithm had the highest accuracy in predicting RP. However, in order to use this algorithm in clinical practice, it is necessary to increase data diversity and the number of patients by sharing data between centers.

Background and Aim: Although the prognosis of nasopharyngeal cancer largely depends on a classification based on the tumor-lymph node metastasis staging system, patients at the same stage may have different clinical outcomes. This study aimed to evaluate the survival prognosis of nasopharyngeal cancer using machine learning. Settings and Design: Original, retrospective. Materials and Methods: A total of 72 patients with a diagnosis of nasopharyngeal cancer who received radiotherapy ± chemotherapy were included in the study. The contribution of patient, tumor, and treatment characteristics to the survival prognosis was evaluated by machine learning using the following techniques: logistic regression, artificial neural network, XGBoost, support-vector clustering, random forest, and Gaussian Naive Bayes. Results: In the analysis of the data set, correlation analysis, and binary logistic regression analyses were applied. Of the 18 independent variables, 10 were found to be effective in predicting nasopharyngeal cancer-related mortality: age, weight loss, initial neutrophil/lymphocyte ratio, initial lactate dehydrogenase, initial hemoglobin, radiotherapy duration, tumor diameter, number of concurrent chemotherapy cycles, and T and N stages. Gaussian Naive Bayes was determined as the best algorithm to evaluate the prognosis of machine learning techniques (accuracy rate: 88%, area under the curve score: 0.91, confidence interval: 0.68-1, sensitivity: 75%, specificity: 100%). Conclusion: Many factors affect prognosis in cancer, and machine learning algorithms can be used to determine which factors have a greater effect on survival prognosis, which then allows further research into these factors. In the current study, Gaussian Naive Bayes was identified as the best algorithm for the evaluation of prognosis of nasopharyngeal cancer.

In this editorial we comment on the article by Chen et al published in the recent issue of the World Journal of Clinical Oncology . Brain metastasis is one of the most serious complications of breast cancer and causes high morbidity and mortality. Brain metastases may involve the brain parenchyma and/or leptomeninges. Symptomatic brain metastases develop in 10%-16% of newly recognized cases each year, and this rate increases to 30% in autopsy series. Depending on the size of the metastatic foci, it may be accompanied by extensive vasogenic edema or may occur as small tumor foci. Since brain metastases are a significant cause of morbidity and mortality, early diagnosis can have significant effects on survival and quality of life. The risk of developing brain metastases emerges progressively due to various patient and tumor characteristics. Patient variability may be particularly important in the susceptibility and distribution of brain metastases because malignant blood must cross the brain barrier and move within the brain parenchyma. Some characteristics of the tumor, such as gene expression, may increase the risk of brain metastasis. Clinical growth, tumor stage, tumor grade, growth receptor positivity, HER2 positivity, molecular subtype (such as triple negative status, luminal/nonluminal feature) increase the risk of developing breast cancer metastasis. Factors related to survival due to breast cancer brain metastasis include both tumor/patient characteristics and treatment characteristics, such as patient age, lung metastasis, surgery for brain metastasis, and HER2 positivity. If cases with a high risk of developing brain metastasis can be identified with the help of clinical procedures and artificial intelligence, survival and quality of life can be increased with early diagnosis and treatment. At the same time, it is important to predict the formation of this group in order to develop new treatment methods in cases with low survival expectancy with brain metastases.

Objective: The study aimed to investigate the relationship between the ratio of lymph node attenuation to primary lesion attenuation on contrast-enhanced CT and the PET/CT standard uptake value (SUVmax) in head and neck squamous cell cancer (HNSCC). Methods: Volunteers with advanced-stage, histopathologically proven HNSCC, indicated to have radiotherapy/chemoradiotherapy, were evaluated for CT and PET/CT for radiotherapy planning. The attenuation and SUVmax of the primary lesion and the largest, possibly metastatic lymph node, and the round index and volume of the lymph node were calculated. The relationship between lymph node/primary lesion attenuation and SUVmax ratios was investigated. The differences in CT findings between the SUVmax < and ≥3 groups were examined. Results: Thirty-two cases with adequate diagnostic quality were studied. There was a very strong positive correlation between the primary lesion and lymph node attenuation (r=0.817, p<0.001), a strong correlation between the lymph node volume and SUVmax (r=0.681, p<0.001), and a moderate negative correlation between lymph node/primary lesion SUVmax and attenuation (r=-0.503, p=0.004). In patients with ≥3 SUVmax, lymph node volume and lymph node/primary lesion SUVmax were significantly higher, and the attenuation ratio was close to 1 (PPV 94.1, 86.3%, respectively). Conclusion: In HNSCC, the lymph node/primary lesion attenuation ratio can be used instead of SUVmax if supported by other conventional CT findings. Metastasis should be considered if lymph node attenuation is similar to primary mass attenuation and excluded if higher. CT attenuation rate can be used as a supportive finding if PET/CT cannot be performed or lymph node SUVmax is close to the acceptable cut-off for metastasis.