Validation of advanced hybrid SPECT/CT system using dynamic anthropomorphic cardiac phantom.

OBJECTIVE: Myocardial blood flow (MBF) assessment can provide incremental diagnostic and prognostic information and thus the validation of dynamic SPECT is of high importance. We recently developed a novel cardiac phantom for dynamic SPECT validation and compared its performance against the GE Discovery NM 530c. We now report its use for validation of a new hybrid SPECT/CT System featuring advanced cadmium zinc telluride (CZT) technology in a ring array detector design (StarGuide™, GE HealthCare). METHODS: Our recently developed cardiac phantom with injected technetium-99m radiotracer was used to create physiological time activity curves (TACs) for the left ventricular (LV) cavity and the myocardium. The TACs allow the calculation of uptake rate (K1) and MBF. The StarGuide system was used to acquire and process the TACs, and these were compared to the TACs produced by the phantom and its mathematical model. Fifteen (15) experiments with different doses representing various MBF values were conducted, and a standard statistic tool was applied for significance. RESULTS: The TACs produced by the StarGuide system had a significant correlation (p < 0.001) with the reference TACs generated by the phantom both for the LV (r = 0.94) and for the myocardium (r = 0.89). The calculated MBF difference between the system and the phantom was 0.14 ± 0.16 ml/min/g and the average relative absolute difference was 13.2 ± 8.1%. A coefficient of variance of ≤ 11% was observed for all MBF subranges. The regional uptake rate values were similar to the global one with a maximum difference of 5%. CONCLUSIONS: Our newly developed dynamic cardiac phantom was used for validation of the dynamic hybrid SPECT/CT CZT-based system (StarGuide™, GE). The accuracy and precision of the system for assessing MBF values were high. The new StarGuide system can reliably perform dynamic SPECT acquisitions over a wide range of myocardial perfusion flow rates.

The use of the reds noninvasive lung fluid monitoring system to assess readiness for discharge in patients hospitalized with acute heart failure: A pilot study.

BACKGROUND: Inadequate decongestion is common in hospitalized heart failure (HF) patients and may contribute to readmissions. Our purpose was to use remote dielectric sensing (ReDS) technology to measure lung congestion at discharge in patients admitted with acute HF and to see if a device-targeted intervention could reduce HF readmission rates. METHODS: We conducted a prospective pilot study of patients admitted with acute decompensated HF randomized to receive standard therapy or ReDS-guided therapy to determine the timing of hospital discharge based on the amount of lung congestion present after diuresis. ReDS measurement was performed for all patients once they were deemed ready for discharge. Patients in the treatment arm with residual lung congestion defined by ReDS ≥39% had HF consultation and further diuresis. RESULTS: Of 108 HF patients (50% male, age 73.6 ±â€¯12.6 years, BMI 29.3 ±â€¯4.3 kg/m2, EF 38.5 ±â€¯15.1%, BNP 1138 ±â€¯987 pg/mL), 32% demonstrated residual lung congestion at the time of proposed hospital discharge. ReDS guided therapy triggered additional diuresis in 30% (18/60) of the patients in the treatment arm (average weight loss 5.6 pounds, p = 0.02). 30-day HF readmission rates were similar in the treatment and the control arms (1.7% vs 4.2%; p = 0.44). Patients discharged as planned with residual lung congestion with ReDS ≥39% had higher 30-day readmission rate compared to patients who were adequately decongested at discharge with ReDS <39% (11.8% vs. 1.4%, p = 0.03). CONCLUSION: In our single-center cohort, ReDS testing demonstrated that 32% of HF patients deemed ready for discharge have clinically significant residual lung congestion which was associated with a higher risk of readmission. ReDS-guided management was associated with significant decongestion but not a reduction in HF readmissions in this sample.

A comparative study for chest radiograph image retrieval using binary texture and deep learning classification.

In this work various approaches are investigated for X-ray image retrieval and specifically chest pathology retrieval. Given a query image taken from a data set of 443 images, the objective is to rank images according to similarity. Different features, including binary features, texture features, and deep learning (CNN) features are examined. In addition, two approaches are investigated for the retrieval task. One approach is based on the distance of image descriptors using the above features (hereon termed the "descriptor"-based approach); the second approach ("classification"-based approach) is based on a probability descriptor, generated by a pair-wise classification of each two classes (pathologies) and their decision values using an SVM classifier. Best results are achieved using deep learning features in a classification scheme.