Reproducibility in Radiomics: A Comparison of Feature Extraction Methods and Two Independent Datasets.

Radiomics involves the extraction of information from medical images that are not visible to the human eye. There is evidence that these features can be used for treatment stratification and outcome prediction. However, there is much discussion about the reproducibility of results between different studies. This paper studies the reproducibility of CT texture features used in radiomics, comparing two feature extraction implementations, namely the MATLAB toolkit and Pyradiomics, when applied to independent datasets of CT scans of patients: (i) the open access RIDER dataset containing a set of repeat CT scans taken 15 min apart for 31 patients (RIDER Scan 1 and Scan 2, respectively) treated for lung cancer; and (ii) the open access HN1 dataset containing 137 patients treated for head and neck cancer. Gross tumor volume (GTV), manually outlined by an experienced observer available on both datasets, was used. The 43 common radiomics features available in MATLAB and Pyradiomics were calculated using two intensity-level quantization methods with and without an intensity threshold. Cases were ranked for each feature for all combinations of quantization parameters, and the Spearman's rank coefficient, rs, calculated. Reproducibility was defined when a highly correlated feature in the RIDER dataset also correlated highly in the HN1 dataset, and vice versa. A total of 29 out of the 43 reported stable features were found to be highly reproducible between MATLAB and Pyradiomics implementations, having a consistently high correlation in rank ordering for RIDER Scan 1 and RIDER Scan 2 (rs > 0.8). 18/43 reported features were common in the RIDER and HN1 datasets, suggesting they may be agnostic to disease site. Useful radiomics features should be selected based on reproducibility. This study identified a set of features that meet this requirement and validated the methodology for evaluating reproducibility between datasets.

The British nuclear medicine research strategy - the framework.

The British Nuclear Medicine Society (BNMS) has developed a Research Strategy framework led by the Research Champions of the BNMS and overseen by the BNMS Research and Innovation Committee. The objectives of the Research Strategy are to improve translation of cutting-edge nuclear medicine research from bench to bedside, the implementation of state-of-the-art multimodality technologies and to enhance multicentre radionuclide research in the UK. It strives to involve patients and the public in radionuclide research and to contribute to and work with the multi-professional national and international organisations involved in research with an ultimate aim to improve nuclear medicine services, and patients' outcomes and care.

The smoking-dyslipidaemia dyad: A potent synergistic risk for atherosclerotic coronary artery disease.

BACKGROUND: Smoking and dyslipidaemia are known individual risk factors of coronary artery disease (CAD). The present study examined the combined risk of smoking and dyslipidaemia on coronary atherosclerosis. METHODS: Coronary artery calcium (CAC), measured by cardiac CT, was used to assess the extent of CAD, which was related to smoking and dyslipidaemia using logistic regression, adjusted for age, sex, hypertension, BMI and family history of ischaemic heart disease. RESULTS: Seventy-one patients (46 men, 25 women: median age of 53.7yrs; IQR = 47.0-59.5) were recruited. The mean log10 CAC score in never-smokers without dyslipidaemia (reference group) was 0.37 (SD = 0.73), while the value in those with a history of smoking was 0.44 ± 0.48 (mean difference: 0.07, 95%CI:-0.67 to 0.81, p = 0.844), dyslipidaemia was 1.07 ± 1.08 (mean difference: 0.71, 95%CI: 0.24 to 1.17, p = 0.003), and both risk factors was 1.82 ± 0.64 (mean difference: 1.45, 95%CI:0.88 to 2.02, p < 0.001). For individuals in the reference group, the proportions with none, one and multiple vessel disease were 80.6%, 16.1% and 3.2%; for those with a history of smoking or with dyslipidaemia were 50.0%, 25.0% and 25.0%; and for those with both risk factors were 8.3%, 25.0% and 66.7%. Patients with a history of both risk factors had greater adjusted risks of having one- vessel disease - OR = 14.3 (95%CI = 2.1-98.2) or multiple vessel disease: OR = 51.8 (95%CI = 4.2-609.6). CONCLUSIONS: Smoking and dyslipidaemia together are associated with high coronary artery calcification and CAD, independent of other major risk factors.

Sex differences in the agreement between left ventricular ejection fraction measured by myocardial perfusion scintigraphy and by echocardiography.

BACKGROUND: Left ventricular ejection fraction (LVEF) is generally measured by echocardiography but is increasingly available with myocardial perfusion scintigraphy. With myocardial perfusion scintigraphy, the threshold of LVEF below which there is a risk for myocardial infarct or sudden cardiac death is higher for women (51%) than for men (43%). We tested the hypothesis that such a sex difference may also occur with echocardiography and myocardial perfusion scintigraphy. METHODS: Four hundred and four men, mean age = 67.7 ± SD = 12.3 yr; 339 women, 67.7 ± 11.7 yr had separate myocardial perfusion scintigraphy and echocardiography examinations within six months. A subset of 327 of these patients (181 men, 68.8 ± 12.1 yr; 146 women, 66.4 ± 12.1 yr) had examinations within one month and were additionally analysed as this sub-group. Myocardial perfusion scintigraphy and echocardiography were used to measure LVEF at rest and their agreement (neither considered as a reference method) was assessed by Bland-Altman plots: LVEF difference (myocardial perfusion scintigraphy minus echocardiography ) against average LVEF ( MPS + Echo 2 ). RESULTS: Of patients who had myocardial perfusion scintigraphy and echocardiography performed within six months, mean LVEF difference = +1.1% (95% limits of agreement: -19.3 to +21.6) in men but +10.9% (-10.7 to +32.5) in women. LVEF difference diverged from zero marginally in men (mean difference = +1.1, 95%CI = +0.1 to +2.1, p = 0.028) but more in women (+10.9, +9.8 to +12.1, p < 0.001). The LVEF difference correlated with average LVEF itself in both men (r = 0.305, p < 0.001) and women (r = 0.361, p < 0.001), and with age in women (r = 0.117, p = 0.031). Similar results were observed for the subset. CONCLUSIONS: Caution should be taken when interpreting LVEF measured by different techniques due to their wide limits of agreement and systematic bias, more markedly in women.

A case of aortitis during cisplatin-based chemotherapy for cervical cancer.

A case of aortitis in a patient undergoing adjuvant cisplatin and topotecan chemotherapy for cervical cancer following presentation with pyrexia of unknown origin and raised inflammatory markers is presented. Although many chemotherapy agents are known to cause small vessel vasculitis and there are several reported cases of large vessel vasculitis following gemcitabine chemotherapy, there is only one previously described case of aortitis following cisplatin administration. This case is presented with corresponding CT and 18F-FDG PET-CT imaging with discussion of the literature regarding vasculitis and chemotherapy.

Textural analysis and lung function study: Predicting lung fitness for radiotherapy from a CT scan.

OBJECTIVE: This study tested the hypothesis that shows advanced image analysis can differentiate fit and unfit patients for radical radiotherapy from standard radiotherapy planning imaging, when compared to formal lung function tests, FEV1 (forced expiratory volume in 1 s) and TLCO (transfer factor of carbon monoxide). METHODS: An apical region of interest (ROI) of lung parenchyma was extracted from a standard radiotherapy planning CT scan. Software using a grey level co-occurrence matrix (GLCM) assigned an entropy score to each voxel, based on its similarity to the voxels around it. RESULTS: Density and entropy scores were compared between a cohort of 29 fit patients (defined as FEV1 and TLCO above 50 % predicted value) and 32 unfit patients (FEV1 or TLCO below 50% predicted). Mean and median density and median entropy were significantly different between fit and unfit patients (p = 0.005, 0.0008 and 0.0418 respectively; two-sided Mann-Whitney test). CONCLUSION: Density and entropy assessment can differentiate between fit and unfit patients for radical radiotherapy, using standard CT imaging. ADVANCES IN KNOWLEDGE: This study shows that a novel assessment can generate further data from standard CT imaging. These data could be combined with existing studies to form a multiorgan patient fitness assessment from a single CT scan.

Clinical applications of textural analysis in non-small cell lung cancer.

Lung cancer is the leading cause of cancer mortality worldwide. Treatment pathways include regular cross-sectional imaging, generating large data sets which present intriguing possibilities for exploitation beyond standard visual interpretation. This additional data mining has been termed "radiomics" and includes semantic and agnostic approaches. Textural analysis (TA) is an example of the latter, and uses a range of mathematically derived features to describe an image or region of an image. Often TA is used to describe a suspected or known tumour. TA is an attractive tool as large existing image sets can be submitted to diverse techniques for data processing, presentation, interpretation and hypothesis testing with annotated clinical outcomes. There is a growing anthology of published data using different TA techniques to differentiate between benign and malignant lung nodules, differentiate tissue subtypes of lung cancer, prognosticate and predict outcome and treatment response, as well as predict treatment side effects and potentially aid radiotherapy planning. The aim of this systematic review is to summarize the current published data and understand the potential future role of TA in managing lung cancer.

Monitoring risk factors of cardiovascular disease in cancer survivors.

There exist published literature for cardiovascular disease (CVD) risk monitoring in cancer survivors but the extent of monitoring in clinical oncology practice is unknown. We performed an interactive survey at a Royal College of Physicians conference (11 November 2016) attended by practitioners with an interest in late effects of cancer treatment and supplemented the survey with an audit among 32 lung cancer survivors treated at St Peter's NHS Hospital in 2012-2016. Among the practitioners, 40% reported CVD risk monitoring performed at least annually, which is compatible with European Group for Blood and Marrow Transplantation Guidelines, but 31% indicated that monitoring was never performed. In contrast, 77% felt that at least an annual assessment was required (p<0.001). Corroborating these data, among the lung cancer survivors, 31% and 16% had lipids or glucose/HbA1C measured annually, and 28% and 31% had never had these tests performed since their cancer treatment. Alerting healthcare providers to review protocols may help reduce CVD after cancer treatments.

Factors influencing the robustness of P-value measurements in CT texture prognosis studies.

Several studies have recently reported on the value of CT texture analysis in predicting survival, although the topic remains controversial, with further validation needed in order to consolidate the evidence base. The aim of this study was to investigate the effect of varying the input parameters in the Kaplan-Meier analysis, to determine whether the resulting P-value can be considered to be a robust indicator of the parameter's prognostic potential. A retrospective analysis of the CT-based normalised entropy of 51 patients with lung cancer was performed and overall survival data for these patients were collected. A normalised entropy cut-off was chosen to split the patient cohort into two groups and log-rank testing was performed to assess the survival difference of the two groups. This was repeated for varying normalised entropy cut-offs and varying follow-up periods. Our findings were also compared with previously published results to assess robustness of this parameter in a multi-centre patient cohort. The P-value was found to be highly sensitive to the choice of cut-off value, with small changes in cut-off producing substantial changes in P. The P-value was also sensitive to follow-up period, with particularly noisy results at short follow-up periods. Using matched conditions to previously published results, a P-value of 0.162 was obtained. Survival analysis results can be highly sensitive to the choice in texture cut-off value in dichotomising patients, which should be taken into account when performing such studies to avoid reporting false positive results. Short follow-up periods also produce unstable results and should therefore be avoided to ensure the results produced are reproducible. Previously published findings that indicated the prognostic value of normalised entropy were not replicated here, but further studies with larger patient numbers would be required to determine the cause of the different outcomes.

Clinical evaluation of reducing acquisition time on single-photon emission computed tomography image quality using proprietary resolution recovery software.

INTRODUCTION: A three-dimensional model-based resolution recovery (RR) reconstruction algorithm that compensates for collimator-detector response, resulting in an improvement in reconstructed spatial resolution and signal-to-noise ratio of single-photon emission computed tomography (SPECT) images, was tested. The software is said to retain image quality even with reduced acquisition time. Clinically, any improvement in patient throughput without loss of quality is to be welcomed. Furthermore, future restrictions in radiotracer supplies may add value to this type of data analysis. AIM: The aims of this study were to assess improvement in image quality using the software and to evaluate the potential of performing reduced time acquisitions for bone and parathyroid SPECT applications. MATERIALS AND METHODS: Data acquisition was performed using the local standard SPECT/CT protocols for 99mTc-hydroxymethylene diphosphonate bone and 99mTc-methoxyisobutylisonitrile parathyroid SPECT imaging. The principal modification applied was the acquisition of an eight-frame gated data set acquired using an ECG simulator with a fixed signal as the trigger. This had the effect of partitioning the data such that the effect of reduced time acquisitions could be assessed without conferring additional scanning time on the patient. The set of summed data sets was then independently reconstructed using the RR software to permit a blinded assessment of the effect of acquired counts upon reconstructed image quality as adjudged by three experienced observers. Data sets reconstructed with the RR software were compared with the local standard processing protocols; filtered back-projection and ordered-subset expectation-maximization. RESULTS: Thirty SPECT studies were assessed (20 bone and 10 parathyroid). The images reconstructed with the RR algorithm showed improved image quality for both full-time and half-time acquisitions over local current processing protocols (P<0.05). CONCLUSION: The RR algorithm improved image quality compared with local processing protocols and has been introduced into routine clinical use. SPECT acquisitions are now acquired at half of the time previously required. The method of binning the data can be applied to any other camera system to evaluate the reduction in acquisition time for similar processes. The potential for dose reduction is also inherent with this approach.

Combined 18F-fluoride and 18F-FDG PET/CT scanning for evaluation of malignancy: results of an international multicenter trial.

UNLABELLED: (18)F-FDG PET/CT is used in a variety of cancers, but because of variable rates of glucose metabolism, not all cancers are reliably identified. (18)F(-) PET/CT allows for the acquisition of highly sensitive and specific images of the skeleton. We prospectively evaluated combined (18)F(-)/(18)F-FDG as a single PET/CT examination for evaluation of cancer patients and compared it with separate (18)F(-) PET/CT and (18)F-FDG PET/CT scans. METHODS: One hundred fifteen participants with cancer were prospectively enrolled in an international multicenter trial evaluating (18)F(-) PET/CT, (18)F-FDG PET/CT, and combined (18)F(-)/(18)F-FDG PET/CT. The 3 PET/CT scans were performed sequentially within 4 wk of one another for each patient. RESULTS: (18)F(-)/(18)F-FDG PET/CT allowed for accurate interpretation of radiotracer uptake outside the skeleton, with findings similar to those of (18)F-FDG PET/CT. In 19 participants, skeletal disease was more extensive on (18)F(-) PET/CT and (18)F(-)/(18)F-FDG PET/CT than on (18)F-FDG PET/CT. In another 29 participants, (18)F(-) PET/CT and (18)F(-)/(18)F-FDG PET/CT showed osseous metastases where (18)F-FDG PET/CT was negative. The extent of skeletal lesions was similar in 18 participants on all 3 scans. CONCLUSION: This trial demonstrated that combined (18)F(-)/(18)F-FDG PET/CT shows promising results when compared with separate (18)F(-) PET/CT and (18)F-FDG PET/CT for evaluation of cancer patients. This result opens the possibility for improved patient care and reduction in health-care costs, as will be further evaluated in future trials.

Efficacy of routine surveillance with positron emission tomography/computed tomography in aggressive non-Hodgkin lymphoma in complete remission: status in a single center.

Post-therapy surveillance imaging in patients with lymphoma remains controversial. We report our experience with positron emission tomography/computed tomography (PET/CT) surveillance in patients with aggressive non-Hodgkin lymphoma in first complete remission (CR). The 138 PET/CTs performed in 52 patients revealed four unsuspected relapses. In one patient, relapse was visualized by fluorodeoxyglucose (FDG) accumulation without any significant CT pathology. The specificity and sensitivity of surveillance PET/CT were 89% and 100%, respectively. The predictive values of positive and negative PET/CTs were 21% and 100%, respectively. The cost of half-yearly routine PET/CT surveillance during the first 2 years in CR was $US8552 per patient and accounted for 81% of the total follow-up costs. PET/CT was effective in detecting unexpected relapse and normal PET/CT supported continuous CR. However, the impact of PET/CT was limited by the high number of false-positive results and PET/CT surveillance was costly compared to CT surveillance.

Pediatric and adolescent lymphoma: comparison of whole-body STIR half-Fourier RARE MR imaging with an enhanced PET/CT reference for initial staging.

PURPOSE: To compare the diagnostic performance of rapid whole-body anatomic magnetic resonance (MR) staging of pediatric and adolescent lymphoma to an enhanced positron emission tomographic (PET)/computed tomographic (CT) reference standard. MATERIALS AND METHODS: Ethical permission was given by the University College London Hospital ethics committee, and informed written consent was obtained from all participants and/or parents or guardians. Thirty-one subjects (age range, 7.3-18.0 years; 18 male, 11 female) with histologically proved lymphoma were prospectively recruited. Pretreatment staging was performed with whole-body short inversion time inversion-recovery (STIR) half-Fourier rapid acquisition with relaxation enhancement (RARE) MR imaging, fluorine 18 fluorodeoxyglucose PET/CT, and contrast agent-enhanced chest CT. Twenty-six subjects had posttreatment PET/CT and compromised our final cohort. Eleven nodal and 11 extranodal sites per patient were assessed on MR imaging by two radiologists in consensus, with a nodal short-axis threshold of >1 cm and predefined extranodal positivity criteria. The same sites were independantly evaluated by two nuclear medicine physicians on PET/CT images. Disease positivity was defined as a maximum standardized uptake value >2.5 or nodal size >1 cm. An unblinded expert panel reevaluated the imaging findings, removing perceptual errors, and derived an enhanced PET/CT reference standard (taking into account chest CT and 3-month follow-up imaging) against which the reported and intrinsic performance of MR imaging was assessed by using the kappa statistic. RESULTS: There was very good agreement between MR imaging and the enhanced PET/CT reference standard for nodal and extranodal staging (kappa = 0.96 and 0.86, respectively) which improved following elimination of perceptual errors (kappa = 0.97 and 0.91, respectively). The sensitivity and specificity of MR imaging (following removal of perceptual error) were 98% and 99%, respectively, for nodal disease and 91% and 99%, respectively, for extranodal disease. CONCLUSION: Whole-body STIR half-Fourier RARE MR imaging of pediatric and adolescent lymphoma can accurately depict nodal and extranodal disease and may provide an alternative nonionizing imaging method for anatomic disease assessment at initial staging.

Quantitative diffusion weighted MRI: a functional biomarker of nodal disease in Hodgkin lymphoma?

PURPOSE: This study explores the relationship between MRI Apparent Diffusion Coefficient (ADC) and PET Standardized Uptake Value (SUV) measurements in pediatric Hodgkin lymphoma. METHODS: Sixteen patients (mean age 15.4 yrs, 8 male) with proven Hodgkin lymphoma were recruited and staged using PET-CT, anatomical MRI and additional 1.5T diffusion weighted imaging (DWI) prior to and following chemotherapy. Pre-treatment lymph nodes and anatomically paired post-treatment residual tissue located on MRI were matched to the corresponding PET-CT. Region of interest (ROI) analysis was used to extract quantitative measurements. Mean ADC (ADC(mean)) and maximum SUV (SUV(max)) were recorded and correlation assessed using Spearman statistics. RESULTS: Fifty-three ROIs were sampled. Pre- and post-treatment ADC(mean) ranged from 0.77 × 10(−3) to 1.79 × 10(−3) (median 1.15 × 10(−3) mm(2)s(−1)) and 1.08 × 10(−3) to 3.18 ×10(−3) (median 1.88 × 10(−3) mm(2)s(−1)), and SUV(max) from 2.60 to 25.4 (median 8.85 mg/ml) and 1.00 to 3.50 mg/ml (median 1.90 mg/ml). Median post-treatment ADC(mean) was higher, and median SUV(max) lower than pretreatment values (p < 0.0001). There was an inverse correlation between pre-treatment ADC(mean) and SUV(max) (p = 0.005) and between fractional change ([post-treatment ­ pre-treatment]/pre-treatment)in ADC(mean) and SUV(max) (p =0.002). CONCLUSION: Our results confirm a strong reciprocal relationship between nodal ADC(mean) and SUV(max) in Hodgkin lymphoma;supporting the potential application of quantitative DWI as a functional biomarker of disease.

Comparison of (68)Ga-DOTATATE and (18)F-fluorodeoxyglucose PET/CT in the detection of recurrent medullary thyroid carcinoma.

PURPOSE: This was a retrospective study to detect and map the extent of disease in recurrent medullary thyroid carcinoma (MTC) using the novel PET somatostatin analogue (68)Ga-DOTATATE and conventional (18)F-FDG positron emission tomography/computed tomography (PET/CT). METHODS: Eighteen patients (13 men, 5 women, median age: 54 years) who had previously been operated on for MTC and presented with biochemical (raised calcitonin levels) and/or imaging evidence of recurrence underwent both (68)Ga-DOTATATE and (18)F-FDG PET/CT within a maximum interval of 4 weeks (median interval of 1 week). (68)Ga-DOTATATE- and (18)F-FDG-avid lesions were recorded per patient as well as per region in six distinct regions: (1) thyroid bed-local recurrence, (2) cervical lymph nodes, (3) mediastinum, (4) lungs, (5) liver and (6) bones. The (68)Ga-DOTATATE and (18)F-FDG PET/CT findings were classified as positive or negative on visual interpretation. These findings were further characterised as concordant or discordant, depending on whether there was agreement or discrepancy in imaging with the two radiotracers. A separate analysis of the unenhanced CT component of the examination was performed. Verification of the lesions was achieved by histopathological analysis, further imaging studies and clinical follow-up. RESULTS: (68)Ga-DOTATATE PET/CT imaging achieved disease detection in 13 of 18 and (18)F-FDG PET/CT in 14 of 18 patients. These results corresponded to per-patient sensitivities of 72.2% [95% confidence interval (CI): 46.4-89.3%] for (68)Ga-DOTATATE versus 77.8% (95% CI: 51.9-92.6%) for (18)F-FDG (non-significant difference). (18)F-FDG revealed a total of 28 metastatic MTC regions and (68)Ga-DOTATATE 23 regions. In ten patients a discordant tracer pattern of per-region and/or per-lesion distribution of recurrent disease was observed, while in four patients a concordant pattern was noted (no lesions were detected by either modality in the remaining four patients). CONCLUSION: Neither (18)F-FDG nor (68)Ga-DOTATATE PET/CT can fully map the extent of disease in patients with recurrent MTC, although (18)F-FDG PET/CT may identify more lesions. However, (68)Ga-DOTATATE PET/CT can be a useful complementary imaging tool and may identify patients suitable for consideration of targeted radionuclide somatostatin analogue therapy.

Attenuation correction for lung SPECT: evidence of need and validation of an attenuation map derived from the emission data.

PURPOSE: The aim of our study was to investigate the importance of attenuation correction (AC) in reconstructed and reprojected images on lung SPECT studies. METHODS: Simulation studies were undertaken to evaluate the influence of AC on defect-to-normal ratios (D/N), to demonstrate the influence of errors in the correction map values and to detect lung boundaries used for AC. The use of a synthetic map (SM) for AC of the clinical data was also evaluated and the results compared with those obtained with data derived from CT (CTM). Additionally, the role of AC in reprojected SPECT data was assessed and level of noise on the 'planar-like' images was measured. RESULTS: Phantom studies showed that AC markedly affects the D/N ratio. However, variations in micro values typical of those found in clinical studies resulted in relatively small changes in results. Eroded and dilated conditions did not cause any significant effect on D/N. The level of noise in the reprojected images is reduced in comparison with real planar data. Clinical SPECT/CT data reconstructed with AC using CTM and SM showed an excellent correlation between the two methods. CONCLUSION: AC improves D/N in lung SPECT studies, thus potentially enhancing the diagnostic capability of the method. The use of a synthetic map for AC is feasible, avoiding the need for an additional procedure and the increased radiation dose involved. Planar-like images generated from reprojected SPECT data are well matched to normal planar images provided AC is performed and attenuation included in the reprojection.