Development and validation of multicentre study on novel Artificial Intelligence-based Cardiovascular Risk Score (AICVD)

OBJECTIVE

Cardiovascular diseases (CVD) are one of the most prevalent diseases in India amounting for nearly 30% of total deaths. A dearth of research on CVD risk scores in Indian population, limited performance of conventional risk scores and inability to reproduce the initial accuracies in randomised clinical trials has led to this study on large-scale patient data. The objective is to develop an Artificial Intelligence-based Risk Score (AICVD) to predict CVD event (eg, acute myocardial infarction/acute coronary syndrome) in the next 10 years and compare the model with the Framingham Heart Risk Score (FHRS) and QRisk3.

METHODS

Our study included 31 599 participants aged 18-91 years from 2009 to 2018 in six Apollo Hospitals in India. A multistep risk factors selection process using Spearman correlation coefficient and propensity score matching yielded 21 risk factors. A deep learning hazards model was built on risk factors to predict event occurrence (classification) and time to event (hazards model) using multilayered neural network. Further, the model was validated with independent retrospective cohorts of participants from India and the Netherlands and compared with FHRS and QRisk3.

RESULTS

The deep learning hazards model had a good performance (area under the curve (AUC) 0.853). Validation and comparative results showed AUCs between 0.84 and 0.92 with better positive likelihood ratio (AICVD -6.16 to FHRS -2.24 and QRisk3 -1.16) and accuracy (AICVD -80.15% to FHRS 59.71% and QRisk3 51.57%). In the Netherlands cohort, AICVD also outperformed the Framingham Heart Risk Model (AUC -0.737 vs 0.707).

CONCLUSIONS

This study concludes that the novel AI-based CVD Risk Score has a higher predictive performance for cardiac events than conventional risk scores in Indian population.

TRIAL REGISTRATION NUMBER

CTRI/2019/07/020471.

Comparison of Bayesian network and support vector machine models for two-year survival prediction in lung cancer patients treated with radiotherapy

PURPOSE

Classic statistical and machine learning models such as support vector machines (SVMs) can be used to predict cancer outcome, but often only perform well if all the input variables are known, which is unlikely in the medical domain. Bayesian network (BN) models have a natural ability to reason under uncertainty and might handle missing data better. In this study, the authors hypothesize that a BN model can predict two-year survival in non-small cell lung cancer (NSCLC) patients as accurately as SVM, but will predict survival more accurately when data are missing.

METHODS

A BN and SVM model were trained on 322 inoperable NSCLC patients treated with radiotherapy from Maastricht and validated in three independent data sets of 35, 47, and 33 patients from Ghent, Leuven, and Toronto. Missing variables occurred in the data set with only 37, 28, and 24 patients having a complete data set.

RESULTS

The BN model structure and parameter learning identified gross tumor volume size, performance status, and number of positive lymph nodes on a PET as prognostic factors for two-year survival. When validated in the full validation set of Ghent, Leuven, and Toronto, the BN model had an AUC of 0.77, 0.72, and 0.70, respectively. A SVM model based on the same variables had an overall worse performance (AUC 0.71, 0.68, and 0.69) especially in the Ghent set, which had the highest percentage of missing the important GTV size data. When only patients with complete data sets were considered, the BN and SVM model performed more alike.

CONCLUSIONS

Within the limitations of this study, the hypothesis is supported that BN models are better at handling missing data than SVM models and are therefore more suitable for the medical domain. Future works have to focus on improving the BN performance by including more patients, more variables, and more diversity.

In vivo dosimetry with a linear MOSFET array to evaluate the urethra dose during permanent implant brachytherapy using iodine-125

PURPOSE

To develop a technique to monitor the dose rate in the urethra during permanent implant brachytherapy using a linear MOSFET array, with sufficient accuracy and without significantly extending the implantation time.

METHODS AND MATERIALS

Phantom measurements were performed to determine the optimal conditions for clinical measurements. In vivo measurements were performed in 5 patients during the (125)I brachytherapy implant procedure. To evaluate if the urethra dose obtained in the operating room with the ultrasound transducer in the rectum and the patient in treatment position is a reference for the total accumulated dose; additional measurements were performed after the implantation procedure, in the recovery room.

RESULTS

In vivo measurements during and after the implantation procedure agree very well, illustrating that the ultrasound transducer in the rectum and patient positioning do not influence the measured dose in the urethra. In vivo dose values obtained during the implantation are therefore representative for the total accumulated dose in the urethra. In 5 patients, the dose rates during and after the implantation were below the maximum dose rate of the urethra, using the planned seed distribution.

CONCLUSION

In vivo dosimetry during the implantation, using a MOSFET array, is a feasible technique to evaluate the dose in the urethra during the implantation of (125)I seeds for prostate brachytherapy.

Identification of residual metabolic-active areas within individual NSCLC tumours using a pre-radiotherapy (18)Fluorodeoxyglucose-PET-CT scan

BACKGROUND AND PURPOSE

Non-small cell lung cancer (NSCLC) tumours are mostly heterogeneous. We hypothesized that areas within the tumour with a high pre-radiation (18)F-deoxyglucose (FDG) uptake, could identify residual metabolic-active areas, ultimately enabling selective-boosting of tumour sub-volumes.

MATERIAL AND METHODS

Fifty-five patients with inoperable stage I-III NSCLC treated with chemo-radiation or with radiotherapy alone were included. For each patient one pre-radiotherapy and one post-radiotherapy FDG-PET-CT scans were available. Twenty-two patients showing persistent FDG uptake in the primary tumour after radiotherapy were analyzed. Overlap fractions (OFs) were calculated between standardized uptake value (SUV) threshold-based auto-delineations on the pre- and post-radiotherapy scan.

RESULTS

Patients with residual metabolic-active areas within the tumour had a significantly worse survival compared to individuals with a complete metabolic response (p=0.002). The residual metabolic-active areas within the tumour largely corresponded (OF>70%) with the 50%SUV high FDG uptake area of the pre-radiotherapy scan. The hotspot within the residual area (90%SUV) was completely within the GTV (OF=100%), and had a high overlap with the pre-radiotherapy 50%SUV threshold (OF>84%).

CONCLUSIONS

The location of residual metabolic-active areas within the primary tumour after therapy corresponded with the original high FDG uptake areas pre-radiotherapy. Therefore, a single pre-treatment FDG-PET-CT scan allows for the identification of residual metabolic-active areas.

[Protons and ions in the treatment of cancer; a systematic review of the literature]

OBJECTIVE

To provide an overview of the present role of proton and ion therapy, also referred to as 'charged particle therapy', in the treatment of cancer.

DESIGN

Systematic literature study.

METHOD

Systematic electronic searches were carried out in 12 databases according to the Cochrane Collaboration criteria, without restriction as to year of publication or study design. Manual searches of bibliographies and journals were also performed. The inclusion criteria were: at least 20 patients and a follow-up of at least 2 years. In addition, experts on the subject were consulted by correspondence for their opinion.

RESULTS

The search identified 36 relevant articles on proton therapy and 15 on ion therapy. Based on prospective and retrospective studies, proton irradiation emerged as the treatment of choice for ocular tumours, chordomas and skull-base tumours. For prostate cancer, the results were comparable with the best results of photon therapy. Ion therapy was still in an experimental phase.

CONCLUSION

According to the current literature, proton therapy is looked upon as the preferred treatment modality for certain rare tumours, such as ocular tumours, chordoma, and skull-base tumours. However, charged particle therapy as a whole, and especially ion therapy, is not supported as the treatment of choice for cancer by published evidence. Nevertheless, the potential theoretical benefit of this treatment is great.

Phased attenuation correction in respiration correlated computed tomography/positron emitted tomography

The motion of lung tumors with respiration causes difficulties in the imaging with computed tomography (CT) and positronemitted tomography (PET). Since an accurate knowledge of the position of the tumor and the surrounding tissues is needed for radiation treatment planning, it is important to improve CT/PET image acquisition. The purpose of this study was to evaluate the potential to improve image acquisition using phased attenuation correction in respiration correlated CT/PET, where data of both modalities were binned retrospectively. Respiration correlated scans were made on a Siemens Biograph Sensation 16 CT/PET scanner which was modified to make a low pitch CT scan and list mode PET scan possible. A lollipop phantom was used in the experiments. The sphere with a diameter of 3.1 cm was filled with approximately 20 MBq 18F-FDG. Three longitudinal movement amplitudes were tested: 2.5, 3.9, and 4.8 cm. After collection of the raw CT data, list mode PET data, and the respiratory signal CT/PET images were binned to ten phases with the help of in-house-built software. Each PET phase was corrected for attenuation with CT data of the corresponding phase. For comparison, the attenuation correction was also performed with nonrespiration correlated (non-RC) CT data. The volume and the amplitude of the movement were calculated for every phaseof both the CT and PET data (with phased attenuation correction). Maximum and average activity concentrations were compared between the phased and nonphased attenuation corrected PET. With a standard non-RC CT/PET scan, the volume was underestimated by as much as 46% in CT and the PET volume was overestimated to 370%. The volumes found with RC-CT/PET scanning had average deviations of 1.9% (+/- 4.8%) and 1.5% (+/- 3.4%) from the actual volume, for the CT and PET volumes, respectively. Evaluation of the maximum activity concentration showed a clear displacement in the images with non-RC attenuation correction, and activity values were on average14% (+/- 12%) lower than with phased attenuation correction. The standard deviation of the maximum activity values found in the different phases was a factor of 10 smaller when phased attenuation correction was applied. In this phantom study, we have shown that a combination of respiration correlated CT/PET scanning with application of phased attenuation correction can improve the imaging of moving objects and can lead to improved volume estimation and a more precise localization and quantification of the activity.

Minimal invasive epicardial lead implantation: optimizing cardiac resynchronization with a new mapping device for epicardial lead placement

To optimize resynchronization in biventricular pacing with epicardial leads, mapping to determine the best pacing site, is a prerequisite. A port access surgical mapping technique was developed that allowed multiple pace site selection and reproducible lead evaluation and implantation. Pressure-volume loops analysis was used for real time guidance in targeting epicardial lead placement. Even the smallest changes in lead position revealed significantly different functional results. Optimizing the pacing site with this technique allowed functional improvement up to 40% versus random pace site selection.

Epicardial left ventricular lead placement for cardiac resynchronization therapy: optimal pace site selection with pressure-volume loops

OBJECTIVES

Patients in heart failure with left bundle branch block benefit from cardiac resynchronization therapy. Usually the left ventricular pacing lead is placed by coronary sinus catheterization; however, this procedure is not always successful, and patients may be referred for surgical epicardial lead placement. The objective of this study was to develop a method to guide epicardial lead placement in cardiac resynchronization therapy.

METHODS

Eleven patients in heart failure who were eligible for cardiac resynchronization therapy were referred for surgery because of failed coronary sinus left ventricular lead implantation. Minithoracotomy or thoracoscopy was performed, and a temporary epicardial electrode was used for biventricular pacing at various sites on the left ventricle. Pressure-volume loops with the conductance catheter were used to select the best site for each individual patient.

RESULTS

Relative to the baseline situation, biventricular pacing with an optimal left ventricular lead position significantly increased stroke volume (+39%, P =.01), maximal left ventricular pressure derivative (+20%, P =.02), ejection fraction (+30%, P =.007), and stroke work (+66%, P =.006) and reduced end-systolic volume (-6%, P =.04). In contrast, biventricular pacing at a suboptimal site did not significantly change left ventricular function and even worsened it in some cases.

CONCLUSIONS

To optimize cardiac resynchronization therapy with epicardial leads, mapping to determine the best pace site is a prerequisite. Pressure-volume loops offer real-time guidance for targeting epicardial lead placement during minimal invasive surgery.