Primary hyperparathyroidism, a machine learning approach to identify multiglandular disease in patients with a single adenoma found at preoperative Sestamibi-SPECT/CT.

Objective: Successful preoperative image localisation of all parathyroid adenomas (PTA) in patients with primary hyperparathyroidism (pHPT) and multiglandular disease (MGD) remains challenging. We investigate whether a machine learning classifier (MLC) could predict the presence of overlooked PTA at preoperative localisation with 99mTc-Sestamibi-SPECT/CT in MGD patients. Design: This study is a retrospective study from a single tertiary referral hospital initially including 349 patients with biochemically confirmed pHPT and cured after surgical parathyroidectomy. Methods: A classification ensemble of decision trees with Bayesian hyperparameter optimisation and five-fold cross-validation was trained with six predictor variables: the preoperative plasma concentrations of parathyroid hormone, total calcium and thyroid-stimulating hormone, the serum concentration of ionised calcium, the 24-h urine calcium and the histopathological weight of the localised PTA at imaging. Two response classes were defined: patients with single-gland disease (SGD) correctly localised at imaging and MGD patients in whom only one PTA was localised on imaging. The data set was split into 70% for training and 30% for testing. The MLC was also tested on a subset of the original data based on CT image-derived PTA weights. Results: The MLC achieved an overall accuracy at validation of 90% with an area under the cross-validation receiver operating characteristic curve of 0.9. On test data, the MLC reached a 72% true-positive prediction rate for MGD patients and a misclassification rate of 6% for SGD patients. Similar results were obtained in the testing set with image-derived PTA weight. Conclusions: Artificial intelligence can aid in identifying patients with MGD for whom 99mTc-Sestamibi-SPECT/CT failed to visualise all PTAs.

The preoperative localisation of small parathyroid adenomas improves when adding Tc-99m-Sestamibi SPECT to multiphase contrast-enhanced CT.

OBJECTIVES: To investigate the incremental value of Sestamibi SPECT combined with a non-enhanced and contrast-enhanced CT, using SPECT/CT, for the preoperative localisation of small parathyroid adenomas (PTA). METHODS: Retrospectively, 147 patients surgically cured from primary hyperparathyroidism, as verified by biochemistry 6 months postoperatively, were included. All patients had preoperatively undergone a dual time 99mTechnetium-Sestamibi SPECT (S) with multiphase CT including native (N), arterial (A) and venous (V) phases. Independently, two radiologists blinded from both the surgical and the preoperative imaging reports, sequentially performed PTA localisation starting with either [A] or [V], thereafter [A + N] or [V + N] and finally with the complete [A + N + S] or [V + N + S]. PTA localisation was reported for each image-set. The readers results were combined and the diagnostic performance for each image set was determined. Sensitivity was also calculated for the different quartiles of PTA weight distribution. RESULTS: The median adenoma weight was 315 mg. No statistically significant differences in diagnostic performance between arterial and venous based image sets were found. The net effect of adding [N] was to increase specificity. Sestamibi SPECT significantly increased the overall diagnostic accuracy for arterial- and venous-based image sets, p = 0.0008 and p = 0.001, respectively. [A + N + S] was found to have the highest diagnostic performance with 86.5% sensitivity and 94.9% overall accuracy. [A + N + S] was particularly advantageous for locating PTA in the lower weight quartiles. CONCLUSIONS: Native CT-phase and dual time point Sestamibi SPECT increase specificity and sensitivity, respectively. These, in combination with a single contrast-enhanced CT-phase is the most optimal examination protocol for preoperative localisation of PTA using SPECT/CT.

Multiphase Iodine Contrast-Enhanced SPECT/CT Outperforms Nonenhanced SPECT/CT for Preoperative Localization of Small Parathyroid Adenomas.

PURPOSE: The aim of this study was to assess the value of intravenously contrast-enhanced CT in conjunction with Tc-MIBI SPECT for preoperative localization of parathyroid adenoma. METHODS: One hundred ninety-two patients with primary hyperparathyroidism were enrolled in the study between May 2015 and May 2017. The patients underwent a preoperative "one-stop shop" examination with Tc-MIBI SPECT/CT by using dual time-point (10 and 90 minutes) protocol and both nonenhanced CT and contrast-enhanced CT acquisition in the arterial and venous phase, 35 and 75 seconds, respectively, after contrast medium injection start. For 149 patients, the imaging results could be correlated to those at surgery and histopathology. RESULTS: The median adenoma weight was 330 mg. The addition of contrast-enhanced CT increased the sensitivity from 81.1% to 89.9% (P = 0.003). The specificity of nonenhanced SPECT/CT was similar to contrast-enhanced CT (96.1% vs 97.9%; P = 0.077). For patients with uniglandular disease (n = 140, 94.0%), the sensitivity increased from 86.4% to 93.6% (P = 0.021) and the specificity from 96.2% to 97.9% (P = 0.118) by adding contrast-enhanced CT. In patients with multiglandular disease (n = 9, 6.0%), adding contrast-enhanced CT improved detection sensitivity from 42.1% to 63.2%. However, these patients were few and significance was not reached (P = 0.125). CONCLUSIONS: In this cohort, with generally small parathyroid adenomas, the sensitivity in preoperative localization was greatly improved by adding contrast-enhanced CT to Tc-MIBI SPECT/CT.

Comparison of acquisition protocols for ventilation/perfusion SPECT-a Monte Carlo study.

One of the most commonly used imaging techniques for diagnosing pulmonary embolism (PE) is ventilation/perfusion (V/P) scintigraphy. The aim of this study was to evaluate the performance of the currently used imaging protocols for V/P single photon emission computed tomography (V/P SPECT) at two nuclear medicine department sites and to investigate the effect of altering important protocol parameters. The Monte Carlo technique was used to simulate 4D digital phantoms with perfusion defects. Six imaging protocols were included in the study and a total of 72 digital patients were simulated. Six dually trained radiologists/nuclear medicine physicians reviewed the images and reported all perfusion mismatch findings. The radiologists also visually graded the image quality. No statistically significant differences in diagnostic performance were found between the studied protocols, but visual grading analysis pointed out one protocol as significantly superior to four of the other protocols. Considering the study results, we have decided to harmonize our clinical protocols for imaging patients with suspected PE. The administered Technegas and macro aggregated albumin activities have been altered, a low energy all purpose collimator is used instead of a low energy high resolution collimator and the acquisition times have been lowered.

SPECT/CT's Advantage for Preoperative Localization of Small Parathyroid Adenomas in Primary Hyperparathyroidism.

PURPOSES: The aims of this study were to assess the performance of Tc-sestamibi SPECT/CT, with diagnostic CT quality, compared with SPECT alone for preoperative localization of parathyroid adenomas and to assess the influence of adenoma weight on the correct adenoma lateralization with SPECT/CT and with SPECT alone. METHODS: Two hundred forty-nine consecutive patients, biochemically diagnosed with primary hyperparathyroidism, were examined with a combined SPECT/CT scanner. Subsequently, 200 patients with confirmed histopathology and biochemical cure after parathyroidectomy were included in this study (16 with multiglandular disease). For each patient, the SPECT-alone data were analyzed first. Thereafter, the CT information was added, and a new evaluation was performed with the combined data. In addition, for each patient, the diagnostic confidence with each method was graded on a scale based on the presence of different image features. The preoperative diagnostic findings were then compared with the surgical and histopathologic reports. RESULTS: The distribution of adenoma weights showed a peak at 210 mg, with a median at 338 mg. The sensitivity and specificity (multiglandular disease included) for correct classification of adenomas were significantly higher for SPECT/CT, 83% and 96%, respectively, than for SPECT alone, 80% and 93% (P < 0.01). Below 210 mg, the differences between SPECT/CT and SPECT alone in accurate adenoma lateralization were more prominent. Sixty-seven percent of all adenomas were graded with the highest confidence score with SPECT/CT compared with 53% with SPECT. CONCLUSIONS: SPECT/CT yields fewer false-positive findings than SPECT alone. The advantage of SPECT/CT over SPECT alone was most apparent for correct lateralization of small adenomas (<210 mg).

Reduced acquisition times in whole body bone scintigraphy using a noise-reducing Pixon®-algorithm-a qualitative evaluation study.

BACKGROUND: Reducing scan-time while maintaining sufficient image quality is a common issue in nuclear medicine diagnostics. This matter can be addressed by different post-processing methods such as Pixon® image processing. The aim of the present study was to evaluate if a commercially available noise-reducing Pixon-algorithm applied on whole body bone scintigraphy acquired with half the standard scan-time could provide the same clinical information as full scan-time non-processed images. METHODS: Twenty patients were administered with 500 MBq (99m)Tc-diphosphonate and scanned on a Siemens Symbia T16 system. Each patient was first imaged using a standard clinical protocol and subsequently imaged using a protocol with half the standard scan-time. Half-time images were processed using a commercially available software package, Enhanced Planar Processing, from Siemens. All images were anonymized and visually evaluated with regard to clinically relevant lesion detectability by three experienced nuclear medicine physicians. The result of this evaluation was grouped into four BMI intervals to investigate the performance of the algorithm with regard to different patient size. Also, a comparison study was performed where the physicians compared the standard image and the processed half-time image corresponding to the same patient with regard to lesion detectability, image noise, and artifacts. RESULTS: The results showed that 93 % of the processed half-time images and 98 % of the standard images were rated as sufficient or good with regard to lesion detectability. The processed half-time images were predominately considered sufficient (65 %), whereas the majority of the standard images were graded as good (83 %). The performance of the algorithm was unaffected by patient size as the average grading of all half-time processed images was constant independent of patient BMI. The comparison study showed that the standard images were rated superior with regard to lesion detectability, image noise, and artifacts, in 32, 65, and 23 % of the evaluations, respectively. CONCLUSIONS: The results indicate that the Pixon Enhanced Planar Processing does not fully compensate for the loss of counts associated with reducing the scan-time in half for whole body bone scintigraphies. The findings showed that implementing the Pixon-algorithm on images acquired with half the acquisition time in overall provide sufficient clinical information regardless of patient size. The half-time processed images were predominantly graded lower in comparison to images acquired with full time protocols, and a less aggressive reduction in scan-time is therefore recommended.

Gene expression profiling of sequential metastatic biopsies for biomarker discovery in breast cancer.

The feasibility of longitudinal metastatic biopsies for gene expression profiling in breast cancer is unexplored. Dynamic changes in gene expression can potentially predict efficacy of targeted cancer drugs. Patients enrolled in a phase III trial of metastatic breast cancer with docetaxel monotherapy versus combination of docetaxel + sunitinib were offered to participate in a translational substudy comprising longitudinal fine needle aspiration biopsies and Positron Emission Tomography imaging before (T1) and two weeks after start of treatment (T2). Aspirated tumor material was used for microarray analysis, and treatment-induced changes (T2 versus T1) in gene expression and standardized uptake values (SUV) were investigated and correlated to clinical outcome measures. Gene expression profiling yielded high-quality data at both time points in 14/18 patients. Unsupervised clustering revealed specific patterns of changes caused by monotherapy vs. combination therapy (p = 0.021, Fisher's exact test). A therapy-induced reduction of known proliferation and hypoxia metagene scores was prominent in the combination arm. Changes in a previously reported hypoxia metagene score were strongly correlated to the objective responses seen by conventional radiology assessments after 6 weeks in the combination arm, Spearman's ρ = 1 (p = 0.017) but not in monotherapy, ρ = -0.029 (p = 1). Similarly, the Predictor Analysis of Microarrays 50 (PAM50) proliferation metagene correlated to tumor changes merely in the combination arm at 6 and 12 weeks (ρ = 0.900, p = 0.083 and ρ = 1, p = 0.017 respectively). Reductions in mean SUV were a reliable early predictor of objective response in monotherapy, ρ = 0.833 (p = 0.008), but not in the combination arm ρ = -0.029 (p = 1). Gene expression profiling of longitudinal metastatic aspiration biopsies was feasible, demonstrated biological validity and provided predictive information.