Applications of Generative Adversarial Networks (GANs) in Positron Emission Tomography (PET) imaging: A review.

PURPOSE: This paper reviews recent applications of Generative Adversarial Networks (GANs) in Positron Emission Tomography (PET) imaging. Recent advances in Deep Learning (DL) and GANs catalysed the research of their applications in medical imaging modalities. As a result, several unique GAN topologies have emerged and been assessed in an experimental environment over the last two years. METHODS: The present work extensively describes GAN architectures and their applications in PET imaging. The identification of relevant publications was performed via approved publication indexing websites and repositories. Web of Science, Scopus, and Google Scholar were the major sources of information. RESULTS: The research identified a hundred articles that address PET imaging applications such as attenuation correction, de-noising, scatter correction, removal of artefacts, image fusion, high-dose image estimation, super-resolution, segmentation, and cross-modality synthesis. These applications are presented and accompanied by the corresponding research works. CONCLUSION: GANs are rapidly employed in PET imaging tasks. However, specific limitations must be eliminated to reach their full potential and gain the medical community's trust in everyday clinical practice.

Comparison of renal parenchymal trauma after standard, mini and ultra-mini percutaneous tract dilation in porcine models.

OBJECTIVE: To evaluate whether reducing tract dilation diameter in PCNL (percutaneous nephrolithotomy) procedures results in minimizing of renal trauma of the percutaneous tract. METHODS: A percutaneous renal access tract was established bilaterally to 11 pigs. Two pigs were euthanized immediately after the experiment, while nine pigs were sacrificed 1 month later. The percutaneous accesses were dilated up to 30Fr, 22Fr or 12Fr. The animals underwent a contrast-enhanced computer tomography immediately after the procedure and 30 days later. DMSA-scintigraphy with SPECT-CT was also performed. The kidneys of all animals were harvested for histological evaluation. The volume of scar tissue and the percentage of renal volume replaced by scar tissue were calculated. RESULTS: Immediate post-procedural CT-scans revealed a significant difference in defect diameter among the three modalities. However, the scar volume calculated on CT-images and histopathology showed a significant difference only when 30Fr dilation was compared to 12Fr dilation. The percentage of scar volume was negligible in all cases, but there was still a statistical difference between 30 and 12Fr dilation. Dilation up to 22Fr revealed no statistical differences compared to the other two modalities. DMSA-scintigraphy showed no scar tissue in any case. CONCLUSION: Dilation up to 30Fr may cause a significantly larger scar tissue on renal parenchyma compared to 12Fr dilation as it was shown on CT-images and microscopic evaluation, but based on the DMSA/SPECT-CT this difference seems to be insignificant to the renal function. The scar tissue caused by 22Fr dilation seemed to have no significant difference from the other modalities.

Automatic characterization of myocardial perfusion imaging polar maps employing deep learning and data augmentation.

OBJECTIVE: To investigate a deep learning technique, more specifically state-of-the-art convolutional neural networks (CNN), for automatic characterization of polar maps derived from myocardial perfusion imaging (MPI) studies for the diagnosis of coronary artery disease. SUBJECTS AND METHODS: Stress and rest polar maps corresponding to 216 patient cases from the database of the department of Nuclear Medicine of our institution were analyzed. Both attenuation-corrected (AC) and non-corrected (NAC) images were included. All patients were subjected to invasive coronary angiography within 60 days from MPI. As the initial dataset of this study was small to train a deep learning model from scratch, two strategies were followed. The first is called transfer learning. For this, we employed the state-of-the-art CNN called VGG16, which has been broadly exploited in medical imaging classification tasks. The second strategy involves data augmentation, which is achieved by the rotation of the polar maps, to expand the training set. We evaluated VGG16 with 10-fold cross-validation on the original set of images performing separate experiments for AC and NAC polar maps, as well as for their combination. The results were compared to the standard semi-quantitative polar map analysis based on summed stress and summed difference scores, as well as to the medical experts' diagnostic yield. RESULTS: With reference to the findings of coronary angiography, VGG16 achieved an accuracy of 74.53%, sensitivity 75.00% and specificity 73.43% when the AC and NAC polar maps were incorporated into one single image set. Respective figures of MPI interpretation by experienced Nuclear Medicine physicians were 75.00%, 76.97% and 70.31%. The accuracy of semi-quantitative polar map analysis was lower, 66.20% and 64.81% for AC and NAC technique, respectively. CONCLUSION: The proposed deep learning model with data augmentation techniques performed better than the conventional semi-quantitative polar map analysis and competed with doctor's expertise in this particular patient cohort and image set. The model could potentially serve as an assisting tool to support interpretation of MPI studies or could be used for teaching purposes.

90Y-microsphere radioembolization: The method, clinical evidence and perspective.

The current review unfolds the procedural steps and the clinical evidence for yttrium-90 (90Y)-microspheres radioembolization. Radioembolization is part of the loco-regional therapeutic spectrum for liver malignancy and involves the invasive, intra-arterial delivery of microspheres carrying ß-emitter isotopes in order to destroy cancerous tissue via ionizing radiation. The main steps of the therapeutic process are selection of eligible patients, angiographic workup, simulation scintigraphy, pre-treatment dosimetry, actual treatment and post-treatment imaging/dosimetry. Radioembolization is routinely applied in advanced stage hepatocellular carcinoma (HCC), yet its role is being investigated even in earlier stages. Prospective, randomized controlled trials did not verify increased overall survival of radioembolization over systemic treatment with sorafenib in HCC; however, it showed survival benefit in certain sub-groups and a favorable toxicity profile with fewer adverse events. Radioembolization is also applied in metastatic colon cancer showing tumoral liver responses, which however did not translate into an overall survival benefit. Data regarding applications of this method in other neoplasms, such as neuroendocrine tumors, breast cancer and melanoma are also presented. There are ongoing clinical trials to define the role of radioembolization within recent treatments algorithms, to determine optimal combinations of this treatment with systemic and targeted therapies and to decide the patients' sub-groups, who will benefit the most.

Robust regional cerebral blood flow perfusion deficits in relapsing-remitting multiple sclerosis patients with executive function impairment.

OBJECTIVES: Cognitive impairment is present in up to 65% of Relapsing Remitting Multiple sclerosis (RRMS) patients and can be extremely debilitating. Although deficits in episodic memory and processing speed occur more frequently than executive deficits, executive dysfunction tends to have a significant impact on MS patients' ability to generate strategies, think divergently, solve and estimate problems, and reason in abstract terms with substantial negative impacts on activities of daily living. In the present study we investigated perfusion detection rate and pattern, as well as the association between perfusion rates and cognitive dysfunction in cognitively impaired RRMS patients. METHODS: We present findings from 17 cognitively impaired RRMS patients who were evaluated with a comprehensive neuropsychological battery and additionally evaluated by brain perfusion radiopharmaceutical technetium-99m hexamethyl-propylene amine oxime (99mTc HMPAO). RESULTS: RRMS patients had hypoperfusion in several predefined Brodmann areas and lobes of the brain, relatively to demographically matched healthy controls according to an established normative database NeuroGam™. However, we noted blood flow reduction, mainly in the frontal lobes and other related prefrontal areas, involving both hemispheres, but with asymmetric left hemisphere predominance. Moreover, associations between measures of response inhibition, set shifting (executive functions) and severity of hypoperfusion in the left frontal lobes were also established. CONCLUSION: Cerebral hypoperfusion is an integral feature of MS pathology. Executive dysfunction is associated with robust cerebral perfusion deficits in the frontal and prefrontal cortex of cognitively impaired RRMS patients.

The impact of transmission-emission misregistration on the interpretation of SPET/CT myocardial perfusion studies and the value of misregistration correction.

OBJECTIVE: Previous studies indicate that the quality of single photon emission tomography/computed tomography (SPET/CT) myocardial perfusion imaging (MPI) is degraded by even mild transmission-emission misregistrations. The purpose of the current study was to investigate the impact of SPET/CT misalignment on the interpretation of MPI and examine the value of a commercial software application for registration correction. SUBJECTS AND METHODS: A total of 255 technetium-99m ((99m)Tc)-tetrofosmin stress/rest MPI examinations in 150 patients were reviewed for SPET/CT misalignment. After registration correction by the software, images were reassessed for interpretation differences from the misregistered study. The diagnostic benefit of reregistration was determined by taking into account the non-attenuation compensated image pattern, combined stress-rest evaluation, gated-SPET data and patient's history. In a phantom experiment and in 3 representative clinical cases, SPET/CT misalignment was purposely created by the software by sequential slice shifts and its effect was evaluated quantitatively. RESULTS: Misregistration ≥1 pixel in at least one direction was observed in 24% of studies. Interpretation of MPI changed after registration correction in 11% of cases with misalignment <1 pixel, in 18% with 1-2 and in 73% with ≥2 pixels. The diagnostic information seemed to improve after registration correction in 58% of studies irrespective of the degree of misregistration. Software-simulated misregistration had dissimilar effects in the phantom and the 3 selected clinical cases. CONCLUSIONS: The impact of SPET/CT misregistration on MPI interpretation although influenced by the degree and direction of slice misplacement, it is also case-specific and hardly predictable. Registration restoration by the software seems worthwhile regardless of misregistration magnitude.

CT-based attenuation correction in Tl-201 myocardial perfusion scintigraphy is less effective than non-corrected SPECT for risk stratification.

BACKGROUND: Previous studies advocate the use of attenuation correction in myocardial perfusion scintigraphy (MPS) for patient risk stratification. METHODS: Six-hundred and thirty-seven unselected patients underwent Tl-201 MPS by a hybrid SPECT/CT system. Attenuation-corrected (AC) and non-corrected (NAC) images were interpreted blindly and summed stress scores (SSS) were calculated. Study endpoints were all-cause mortality and the composites of death/non-fatal acute myocardial infarction (AMI) and death/AMI/late revascularization. RESULTS: During a follow-up of 42.3 ± 12.8 months 24 deaths, 13 AMIs and 28 revascularizations were recorded. SSS groups formed according to event rate distribution across SSS values were: 0-4, 5-13, >13 for NAC and 0-2, 3-9, >9 for AC. Kaplan-Meier functions were statistically significant between NAC SSS groups for all study endpoints. AC discriminated only between SSS 0-2 and >9 for death/AMI and between 0-2 and 3-9 for death/AMI/revascularization. In the univariate Cox regression abnormal NAC (SSS > 4) was accompanied with much higher hazards ratios than abnormal AC (SSS > 2). In the multivariate model abnormal AC yielded no significance for either endpoint whereas abnormal NAC proved independent from other covariates for the composite endpoints. CONCLUSION: Our results challenge the effectiveness of CT-based AC for risk stratification of patients referred for MPS.

Bone scintigraphy depicts bilateral atypical femoral stress fractures with metachronous presentation, long before a complete fracture occurs.

Atypical femoral fractures (AFF), although rare, are recognized more often during the last decade. They are located in the subtrochanteric region or the femoral shaft, may be bilateral, can evolve to complete fractures after bone overload or minimal trauma and have specific radiological features. The complete fractures have horizontal or slightly oblique configuration accompanied by a medial spike, are non-comminuted, and extend to both cortices. There is also generalized cortical thickening of femoral shaft. Newer evidence suggests that AFF are stress or insufficiency fractures, possibly associated with long-term use of bisphoshonates (BP). AFF can also occur in oncologic patients referred for bone scintigraphy and, in such a case, they should be differentiated from bone metastases. We present here a case with bilateral AFF with metachronous appearance in a female patient with a history of breast cancer and osteoporosis. The first AFF had been depicted on bone scintigraphy 3 years before a complete fracture occurred at this site, but the finding was overlooked. A second bone scan performed shortly after the fracture in order to exclude underlying bone metastases disclosed an additional unsuspected incomplete AFF in the contralateral femur, which was confirmed by radiography. In conclusion, oncologists should consider other causes of bone pain besides bone metastatic disease, and physicians interpreting whole body bone scans of oncologic patients should be aware of the entity of AFF, in order to avoid false positive results and provide early information about an impending complete AFF.

Integrating Machine Learning in Clinical Practice for Characterizing the Malignancy of Solitary Pulmonary Nodules in PET/CT Screening.

The study investigates the efficiency of integrating Machine Learning (ML) in clinical practice for diagnosing solitary pulmonary nodules' (SPN) malignancy. Patient data had been recorded in the Department of Nuclear Medicine, University Hospital of Patras, in Greece. A dataset comprising 456 SPN characteristics extracted from CT scans, the SUVmax score from the PET examination, and the ultimate outcome (benign/malignant), determined by patient follow-up or biopsy, was used to build the ML classifier. Two medical experts provided their malignancy likelihood scores, taking into account the patient's clinical condition and without prior knowledge of the true label of the SPN. Incorporating human assessments into ML model training improved diagnostic efficiency by approximately 3%, highlighting the synergistic role of human judgment alongside ML. Under the latter setup, the ML model had an accuracy score of 95.39% (CI 95%: 95.29-95.49%). While ML exhibited swings in probability scores, human readers excelled in discerning ambiguous cases. ML outperformed the best human reader in challenging instances, particularly in SPNs with ambiguous probability grades, showcasing its utility in diagnostic grey zones. The best human reader reached an accuracy of 80% in the grey zone, whilst ML exhibited 89%. The findings underline the collaborative potential of ML and human expertise in enhancing SPN characterization accuracy and confidence, especially in cases where diagnostic certainty is elusive. This study contributes to understanding how integrating ML and human judgement can optimize SPN diagnostic outcomes, ultimately advancing clinical decision-making in PET/CT screenings.

PDE5 inhibition against acute renal ischemia reperfusion injury in rats: does vardenafil offer protection?

PURPOSE: To evaluate the effect of vardenafil on renal function after renal ischemia-reperfusion (IR) injury (IRI) in a rat model. MATERIALS AND METHODS: Seventy-one Wistar rats were divided into 7 groups including (1) a vehicle-treated group, (2) a vehicle pretreated-IR group, (3-6) vardenafil pretreated-IR groups in doses of 0.02, 0.2, 2 and 20 µg/kg, respectively, (7) a group of IR followed by treatment with 2 µg/kg of vardenafil. Vardenafil or vehicle solution was administered one hour before unilateral nephrectomy and the induction of 45 min of ischemia on the contralateral kidney by clamping of renal pedicle. Four hours of reperfusion were allowed after renal ischemia. Studied parameters were serum creatinine, fractional excretion of sodium (FENa), and histological evaluation of renal specimens. In addition, renal tissue cGMP levels, ERK1/2 phosphorylation as well as renal function by renal scintigraphy were also evaluated. RESULTS: Administration of vardenafil before the induction of ischemia resulted in a significant reduction in creatinine and FENa levels as well as in less histological lesions observed in treated kidneys in comparison with the vehicle-treated group. The underlying mechanism of cytoprotection was cGMP depended and involved the phosphorylation of ERK proteins. Renal scintigraphy confirmed that PDE5 inhibition attenuates renal IRI. CONCLUSIONS: Vardenafil attenuates renal IRI. Based on similar results from relevant studies on other PDE-5 inhibitors in renal and cardiac IRI, it can be assumed that all PDE-5 inhibitors share a common mechanism of cytoprotection.

Is there a role for Nuclear Medicine in diagnosis and management of patients with primary aldosteronism?

Primary aldosteronism (PA) is the most common cause of secondary hypertension. The diagnosis of PA is of clinical importance for choosing the appropriate treatment, meaning, surgery for the unilateral disease, and inclusion of aldosterone antagonists in the antihypertensive treatment for the bilateral disease. Current diagnostic approaches showed that the prevalence of PA is much higher than previously estimated. There is still controversy regarding the true prevalence of PA in hypertensive patients. The gold standard for differentiating between unilateral and bilateral disease is the adrenal vein sampling (AVS), a method that is invasive and is performed accurately in only few dedicated centers. Non invasive methods (imaging) for discriminating the two entities are: the CT scan, MRI and iodocholesterol (NP-59) scintigraphy performed under dexamethasone suppression. But the accuracy of imaging compared to AVS is suboptimal and can result in wrong therapeutic decisions. NP-59 scintigraphy is a non-invasive functional imaging technique that reveals the adrenal cortical autonomic function and could have of incremental value over anatomical imaging. In conclusion, in previous years NP-59 scintigraphy was used infrequently, but recently with the advent of hybrid single photon emission tomography (SPET/CT) systems the interest in NP-59 scintigraphy has been renewed. Studies comparing NP-59 SPET/CT imaging with AVS are warranted in order to establish its diagnostic accuracy.

The effect of percutaneous tract dilation technique on renal parenchymal trauma: An experimental in vivo study on a porcine model.

Purpose: The purpose of this study was to evaluate renal parenchymal trauma of two-step dilation compared to the conventional Amplatz gradual dilation during percutaneous nephrolithotomy on a porcine model. Materials and Methods: A nonpapillary percutaneous access tract was established under fluoroscopic guidance in both kidneys of four female pigs. On the right kidney of each pig, gradual dilation was performed using an Amplatz dilator set with a gradual dilation to 30 Fr, whereas on the left, a two-step dilation was utilized using only 16 Fr and 30 Fr dilators. Two of the animals were euthanized immediately after the procedure and the remaining two 1 month later. The pigs that were kept alive underwent a contrast-enhanced computed tomography immediately, 15, and 30 days postoperatively. A dimercaptosuccinic acid (DMSA) scintigraphy and single-photon emission computed tomography-computed tomography (CT) were also performed after the last CT and afterward, the pigs were sacrificed. All kidneys were harvested for pathohistological examination. Results: The follow-up radiologic imaging showed similar parenchymal damage caused by the compared dilation techniques and an expected reduction in scar size in the later scans. No scar was identified by DMSA in any kidney. Gross and microscopic examinations conducted both on the kidneys that were harvested immediately after the procedure and the ones from the animals that were left to heal, revealed no significant differences in tissue damage, grade of fibrosis, or inflammation depending on the dilation method. Conclusions: Our study showed no inferior outcomes caused by two-step dilation compared to gradual dilation regarding renal parenchymal damage following a nonpapillary puncture. In fact, postoperative imaging findings suggested a trend toward better healing and less scar tissue when the two-step method was used.

ST-segment depression during vasodilator stress is of minor clinical importance in women with normal myocardial perfusion imaging and low or intermediate risk of coronary artery disease.

PURPOSE: The prognostic value of electrocardiographic (ECG) ST-segment depression during vasodilator stress testing in patients with normal myocardial perfusion scintigraphy (MPS) is based on retrospective studies with controversial results. Moreover, the true incidence of obstructive coronary artery disease (CAD) in these patients is unknown. METHODS: During a 33-month period, all consecutive patients referred for MPS were prospectively evaluated for interpretable ST-segment depression ≥ 1 mm during vasodilator stress testing. Of 1,687 patients with normal MPS and without known CAD, 109 (100 women) aged 65.2 ± 9.2 years demonstrated ST-segment shifts and formed the ECG-positive group. The pretest probability of CAD was low in 56%, intermediate in 39% and high in 5%. They were advised to proceed to coronary angiography. An equal number of patients with normal MPS and no ECG abnormalities matched for CAD predisposing factors comprised the control group. Follow-up of both groups was accomplished through a telephone interview. RESULTS: Fifty-two patients from the ECG-positive group (48%) consented to coronary angiography. Six (11.5%) had obstructive CAD (≥50% lumen stenosis), one with left main artery disease, while three required revascularization. Follow-up was complete in 99 ECG-positive patients for a period of 20.6 ± 8.9 months. One hard event (non-fatal myocardial infarction) and one soft event (revascularization) were observed. No event was recorded in the control group. CONCLUSION: In patients with low-intermediate risk of CAD, "ischaemic" ECG changes during vasodilator stress combined with normal MPS are encountered mostly in women and are associated with a low prevalence of significant CAD and a low cardiac event rate.

A Deep Learning Methodology for the Detection of Abnormal Parathyroid Glands via Scintigraphy with 99mTc-Sestamibi.

BACKGROUND: Parathyroid proliferative disorder encompasses a wide spectrum of diseases, including parathyroid adenoma (PTA), parathyroid hyperplasia, and parathyroid carcinoma. Imaging modalities that deliver their results preoperatively help in the localisation of parathyroid glands (PGs) and assist in surgery. Artificial intelligence and, more specifically, image detection methods, can assist medical experts and reduce the workload in their everyday routine. METHODS: The present study employs an innovative CNN topology called ParaNet, to analyse early MIBI, late MIBI, and TcO4 thyroid scan images simultaneously to perform first-level discrimination between patients with abnormal PGs (aPG) and patients with normal PGs (nPG). The study includes 632 parathyroid scans. RESULTS: ParaNet exhibits a top performance, reaching an accuracy of 96.56% in distinguishing between aPG and nPG scans. Its sensitivity and specificity are 96.38% and 97.02%, respectively. PPV and NPV values are 98.76% and 91.57%, respectively. CONCLUSIONS: The proposed network is the first to introduce the automatic discrimination of PG and nPG scans acquired by scintigraphy with 99mTc-sestamibi (MIBI). This methodology could be applied to the everyday routine of medics for real-time evaluation or educational purposes.

Deep learning exploration for SPECT MPI polar map images classification in coronary artery disease.

OBJECTIVE: The exploration and the implementation of a deep learning method using a state-of-the-art convolutional neural network for the classification of polar maps represent myocardial perfusion for the detection of coronary artery disease. SUBJECTS AND METHODS: In the proposed research, the dataset includes stress and rest polar maps in attenuation-corrected (AC) and non-corrected (NAC) format, counting specifically 144 normal and 170 pathological cases. Due to the small number of the dataset, the following methods were implemented: First, transfer learning was conducted using VGG16, which is applied broadly in medical industry. Furthermore, data augmentation was utilized, wherein the images are rotated and flipped for expanding the dataset. Secondly, we evaluated a custom convolutional neural network called RGB CNN, which utilizes fewer parameters and is more lightweight. In addition, we utilized the k-fold validation for evaluating variability and overall performance of the examined model. RESULTS: Our RGB CNN model achieved an agreement rating of 92.07% with a loss of 0.2519. The transfer learning technique (VGG16) attained 95.83% accuracy. CONCLUSIONS: The proposed model could be an effective tool for medical classification problems, in the case of polar map data acquired from myocardial perfusion images.

Parametric images via dynamic 18F-fluorodeoxyglucose positron emission tomographic data acquisition in predicting midterm outcome of liver metastases secondary to gastrointestinal stromal tumours.

PURPOSE: (18)F-Fluorodeoxyglucose positron emission tomography (FDG PET) may underestimate viable tumour tissue in patients with gastrointestinal stromal tumours (GIST) treated with molecular targeted agents. The aim of the present study was to investigate the value of parametric images generated after dynamic data acquisition for the detection of active liver metastases. METHODS: The analysis included 65 dynamic FDG PET studies in 34 patients with liver metastases from GIST who were treated with imatinib or sunitinib. Parametric images of intercept and slope were calculated by dedicated software using a voxel-based linear regression of time-activity data. Intercept images represent the tracer's distribution volume and the slope its overall metabolic turnover. All images were assessed visually and semi-quantitatively. Liver disease status was established 12 months after each PET study. Dichotomous variables of visual interpretation and various quantitative parameters were entered in a statistical model of linear discriminant analysis. RESULTS: Visual analysis of slope images was more sensitive than the standard 1-h FDG uptake evaluation (70.6 vs 51.0%, p = 0.016) in detecting cases with liver disease progression (n = 51). Specificity did not differ. Combination of all variables in the discriminant analysis model correctly classified 87.7% of cases as progressive or non-progressive disease. Sensitivity was raised to 88.2%. CONCLUSION: Parametric images of intercept and slope add a new dimension to the interpretation of FDG PET studies, by isolating visually and quantifying the perfusion and phosphorylation-dependent part of tracer uptake. In treated GIST patients, integration of this information with the 1-h uptake data achieves better characterization of hepatic lesions with respect to disease activity.

Imaging myocardial inflammation of various etiologies with 99mTc-depreotide SPET/CT.

Previous reports suggested the accumulation of technetium-99m-depreotide trifluoroacetate ((99m)Tc-D) at the sites of active infection or inflammation. Binding of depreotide to over-expressed somatostatin receptors in activated lymphocytes and macrophages probably accounts for the depiction of inflammation. We speculated that myocardial inflammation could also be illustrated by (99m)Tc-D scintigraphy. We report on 3 patients with the clinical diagnosis of myocarditis of various etiologies, in which (99m)Tc-D SPET/CT demonstrated obvious tracer uptake in the myocardium of the left ventricle. In conclusion, we suggest that depreotide imaging can depict myocardial inflammation, thus supporting clinical diagnosis.

Advanced fuzzy cognitive maps: state-space and rule-based methodology for coronary artery disease detection.

According to the World Health Organization, 50% of deaths in European Union are caused by Cardiovascular Diseases (CVD), while 80% of premature heart diseases and strokes can be prevented. In this study, a Computer-Aided Diagnostic model for a precise diagnosis of Coronary Artery Disease (CAD) is proposed. The methodology is based on State Space Advanced Fuzzy Cognitive Maps (AFCMs), an evolution of the traditional Fuzzy Cognitive Maps. Also, a rule-based mechanism is incorporated, to further increase the knowledge of the proposed system and the interpretability of the decision mechanism. The proposed method is evaluated utilizing a CAD dataset from the Department of Nuclear Medicine of the University Hospital of Patras, in Greece. Several experiments are conducted to define the optimal parameters of the proposed AFCM. Furthermore, the proposed AFCM is compared with the traditional FCM approach and the literature. The experiments highlight the effectiveness of the AFCM approach, obtaining 85.47% accuracy in CAD diagnosis, showing an improvement of +7% over the traditional approach. It is demonstrated that the AFCM approach in developing Fuzzy Cognitive Maps outperforms the conventional approach, while it constitutes a reliable method for the diagnosis of Coronary Artery Disease.

Multi-input deep learning approach for Cardiovascular Disease diagnosis using Myocardial Perfusion Imaging and clinical data.

PURPOSE: Accurate detection and treatment of Coronary Artery Disease is mainly based on invasive Coronary Angiography, which could be avoided provided that a robust, non-invasive detection methodology emerged. Despite the progress of computational systems, this remains a challenging issue. The present research investigates Machine Learning and Deep Learning methods in competing with the medical experts' diagnostic yield. Although the highly accurate detection of Coronary Artery Disease, even from the experts, is presently implausible, developing Artificial Intelligence models to compete with the human eye and expertise is the first step towards a state-of-the-art Computer-Aided Diagnostic system. METHODS: A set of 566 patient samples is analysed. The dataset contains Polar Maps derived from scintigraphic Myocardial Perfusion Imaging studies, clinical data, and Coronary Angiography results. The latter is considered as reference standard. For the classification of the medical images, the InceptionV3 Convolutional Neural Network is employed, while, for the categorical and continuous features, Neural Networks and Random Forest classifier are proposed. RESULTS: The research suggests that an optimal strategy competing with the medical expert's accuracy involves a hybrid multi-input network composed of InceptionV3 and a Random Forest. This method matches the expert's accuracy, which is 79.15% in the particular dataset. CONCLUSION: Image classification using deep learning methods can cooperate with clinical data classification methods to enhance the robustness of the predicting model, aiming to compete with the medical expert's ability to identify Coronary Artery Disease subjects, from a large scale patient dataset.