Towards survival prediction of cancer patients using medical images.

Survival prediction of a patient is a critical task in clinical medicine for physicians and patients to make an informed decision. Several survival and risk scoring methods have been developed to estimate the survival score of patients using clinical information. For instance, the Global Registry of Acute Coronary Events (GRACE) and Thrombolysis in Myocardial Infarction (TIMI) risk scores are developed for the survival prediction of heart patients. Recently, state-of-the-art medical imaging and analysis techniques have paved the way for survival prediction of cancer patients by understanding key features extracted from Magnetic Resonance Imaging (MRI) and Computed Tomography (CT) scanned images with the help of image processing and machine learning techniques. However, survival prediction is a challenging task due to the complexity in benchmarking of image features, feature selection methods, and machine learning models. In this article, we evaluate the performance of 156 visual features from radiomic and hand-crafted feature classes, six feature selection methods, and 10 machine learning models to benchmark their performance. In addition, MRI scanned Brain Tumor Segmentation (BraTS) and CT scanned non-small cell lung cancer (NSCLC) datasets are used to train classification and regression models. Our results highlight that logistic regression outperforms for the classification with 66 and 54% accuracy for BraTS and NSCLC datasets, respectively. Moreover, our analysis of best-performing features shows that age is a common and significant feature for survival prediction. Also, gray level and shape-based features play a vital role in regression. We believe that the study can be helpful for oncologists, radiologists, and medical imaging researchers to understand and automate the procedure of decision-making and prognosis of cancer patients.

Validity of GRACE Risk Score as a Prognostic Marker of In-hospital Mortality after Acute Coronary Syndrome.

OBJECTIVE: To determine validity of GRACE risk score as a determinant of immediate death during hospitalization for Acute Coronary Syndrome (ACS) and analyze the percentage of cardiac deaths among high, intermediate and low risk groups. STUDY DESIGN: Cross-sectional study. PLACE AND DURATION OF STUDY: Coronary Care Unit of Mayo Hospital, Lahore, from April to July 2015. METHODOLOGY: Patients with acute chest pain were selected according to inclusion and exclusion criteria. Online GRACE risk score calculator was used to determine the predicted risk of death following ACS according to the score. Data was analyzed on SPSS 20. Quantitative data was in the form of median (IQR). Discrimination of GRS was evaluated by using c-statistics, area under the ROC curve. Calibration of GRS was tested by Hosmer-Lameshow test. The correlation between GRACE risk score category and predicted risk of death was determined using Kendall's tau-b bivariate correlation test. Shapiro-Wilk test was applied to check normality of data. The various parameters of GRACE score were studied in patients using Mann-Whitney U-test. The statistically significant p-value was <0.05. RESULTS: There were 165 cases in the study. Overall median GRS was 139 (54). In-hospital deaths were 12.2%. Discrimination of GRS evaluated by area under the ROC curve was 0.913 (95% CI 0.843-0.982; p<0.0001). Application of Hosmer-Lameshow test revealed a p-value of 0.236. Kendall's tau-b bivariate correlation coefficient was 0.384 (p<0.001). CONCLUSION: GRS is an excellent tool to stratify patients of ACS into different risk categories according to various parameters noted at the time of presentation. The risk of predicted death according to the score was variable among different cases, especially those with higher scores. Even though GRS is an effective and valid tool, but it has some tendency of overestimating probability of death following ACS and may require a fine tuning in some cases.

The Trends of CRP Levels at Different Waist-to-Hip Ratios Among Normotensive Overweight and Obese Patients: A Pilot Study.

OBJECTIVE: To determine the correlation between CRP (C-reactive protein) and Waist to Hip Ratio (WHR) among over weight and obese patients with normal blood pressure. STUDY DESIGN: An analytical study. PLACE AND DURATION OF STUDY: Medical indoor and outpatient clinics of Mayo Hospital, Lahore, from March to August 2013. METHODOLOGY: Willing patients with Body Mass Index (BMI) of > 23 kg/m2, normal blood pressures, and age between 18 - 65 years were inducted in the study. Patients with signs of fluid retention, collagen vascular disease, CAD, on corticosteroids, immunomodulators or lipid lowering medications, hypertensives and febrile patients were excluded. Patients were considered to be at low risk for cardiovascular events if WHR among males and females was < 0.95 and < 0.80, respectively. Similarly, males and females with WHR > 1 and > 0.85, respectively were taken as high risk. Levels in-between these ranges were taken as moderate risk. Data was analyzed on SPSS 15. Descriptive statistics were determined. The p-value was calculated by ANOVA and independent sample t-test among males and females respectively, to compare WHR in relation to different CRP levels and < 0.05 was taken as significant. RESULTS: There were 34 male and 74 female patients. The gender-wise mean WHR did not show statistically significant difference categorized CRP levels (p=0.072 in male, and 0.052 in females). There was an increasing trend in CRP levels as WHR increased among females, but this was statistically insignificant (p=0.05). CONCLUSION: Although the impact of central obesity on cardiac health is well known, however, WHR alone is an unreliable indicator of systemic inflammation and raised CRP level.

Correlation of CRP, fasting serum triglycerides and obesity as cardiovascular risk factors.

OBJECTIVE: To determine the correlation of C-reactive protein (CRP) with fasting triglycerides (TG) among pre-obese and obese patients without established diagnosis of coronary artery disease (CAD). STUDY DESIGN: A comparative cross-sectional study. PLACE AND DURATION OF STUDY: Mayo Hospital, Lahore, from January to June 2010. METHODOLOGY: Patients with BMI > 23 kg/m2 aged between 18 - 65 years were inducted and above variables were studied. Patients with signs of fluid retention, collagen vascular disease, CAD, patients on corticosteroids, immunomodulators or lipid lowering medications and febrile patients were not recruited. Body mass index was also determined. Independent sample t-test was applied to see the mean difference of age, CRP level and triglycerides level in relation to gender. Chi-square test was used to see the association between qualitative variables. ANOVA was applied to see CRP and fasting serum TG level in relation to BMI categories. Pearson correlation and simple linear regression was applied to see the dependency of CRP and triglycerides with BMI. P-value ² 0.05 was taken as significant. RESULTS: Raised CRP was major finding among all groups of BMI. Most of obese and pre-obese patients were young and middle aged and belonged to pre-obese group followed by class-1 and class-2 obesity. CRP level increased with body mass index. No such trend was observed for triglycerides. There was an intermediate positive correlation between CRP and BMI and triglycerides and BMI showed a weak negative correlation. If BMI increases by 1 unit on the average, CRP rises by 0.239 times and this unit rise was significant. Whereas 1 unit rise increase in triglycerides on the average cause CRP to decrease -0.006 times but this value was insignificant. CONCLUSION: Raised CRP and high fasting TG were major findings in all age groups especially among young and middle aged people. Obesity, hypertriglyceridemia and raised CRP are interrelated suggesting that obesity is not only linked to hypertriglyceridemia but vascular inflammation among pre-obese and obese without overt diabetes mellitus causes high CRP as well and this can be used as a marker to predict the future risk of CAD. However, in the absence of dyslipidaemia, raised CRP can still be considered as a strong predictor of CAD and stroke.

Concomitant tuberculous meningitis and Lutembacher syndrome with multiple atrial septal defects and infective endocarditis.

Lutembacher syndrome is a rare combination of atrial septal defect (ASD) and mitral stenosis. Symptoms depend on the size of ASD, extent of mitral stenosis and degree of changes in the pulmonary circulation. Presentation can be due to cardiac failure, atrial arrhythmias, dyspnoea, exercise intolerance, paradoxical emboli or other disease related complications like pulmonary hypertension and infective endocarditis. Tuberculous meningitis is a chronic infection due to haematogenous dissemination of tubercle bacilli from lungs. It can lead to complications like cranial nerve palsies, hydrocephalus, cerebral oedema or focal neurological deficits presenting as stroke. The treatment should include antituberculous therapy for one year and corticosteroids for initial 4-6 weeks depending on the symptoms of the patient. This report describes the concomitant occurrence of all these conditions at a time in a 45 years old lady.