Enhanced thyroid nodule segmentation through U-Net and VGG16 fusion with feature engineering: A comprehensive study.

BACKGROUND AND OBJECTIVE: The thyroid gland, a key component of the endocrine system, is pivotal in regulating bodily functions. Thermography, a non-invasive imaging technique utilizing infrared cameras, has emerged as a diagnostic tool for thyroid-related conditions, offering advantages such as early detection and risk stratification. Artificial intelligence (AI) has demonstrated success in medical diagnostics, and its integration into thermal imaging analysis holds promise for improving diagnostic capabilities. This study aims to explore the potential of AI, specifically convolutional neural networks (CNNs), in enhancing the analysis of thyroid thermograms for the detection of nodules and abnormalities. METHODS: Artificial intelligence (AI) and machine learning techniques are integrated to enhance thyroid thermal image analysis. Specifically, a fusion of U-Net and VGG16, combined with feature engineering (FE), is proposed for accurate thyroid nodule segmentation. The novelty of this research lies in leveraging feature engineering in transfer learning for the segmentation of thyroid nodules, even in the presence of a limited dataset. RESULTS: The study presents results from four conducted studies, demonstrating the efficacy of this approach even with a limited dataset. It's observed that in study 4, using FE has led to a significant improvement in the value of the dice coefficient. Even for the small size of the masked region, incorporating radiomics with FE resulted in significant improvements in the segmentation dice coefficient. It's promising that one can achieve higher dice coefficients by employing different models and refining them. CONCLUSION: The findings here underscore the potential of AI for precise and efficient segmentation of thyroid nodules, paving the way for improved thyroid health assessment.

Study of flow, Bioheat transfer and cardiac thermal pulse of aneurysm in the abdominal aortic.

Abdominal aortic aneurysms (AAA) are serious and difficult to detect conditions that can be deadly if they rupture. Infrared thermography (IRT) is a promising imaging technique that can detect abdominal aortic aneurysms more quickly and less costly than other imaging techniques. A clinical biomarker of circular thermal elevation on the midriff skin surface of AAA patient at various scenarios was expected during diagnosis using IRT scanner. However, it is important to note that thermography is not a perfect technology, and it does have some limitations, such as lack of clinical trials. There is still work to be done to improve this imaging technique and make it a more viable and accurate method in detecting abdominal aortic aneurysms. Nevertheless, thermography is currently one of the most convenient technologies in imaging, and it has the potential to detect abdominal aortic aneurysms earlier than other techniques. Cardiac thermal pulse (CTP), on the other hand, was used to examine the thermal physics of AAA. AAA had a CTP that only responded to systolic phase at regular body temperature. Whereas the AAA wall would establish thermal homeostasis with blood temperature following a quasi-linear relationship as the body experienced fever or stage-2 hypothermia. In contrast, a healthy abdominal aorta displayed a CTP that responded to the full cardiac cycle, including diastolic phase at all simulated scenarios.

CFD Computation of Flow Fractional Reserve (FFR) in Coronary Artery Trees Using a Novel Physiologically Based Algorithm (PBA) Under 3D Steady and Pulsatile Flow Conditions.

A novel physiologically based algorithm (PBA) for the computation of fractional flow reserve (FFR) in coronary artery trees (CATs) using computational fluid dynamics (CFD) is proposed and developed. The PBA was based on an extension of Murray's law and additional inlet conditions prescribed iteratively and was implemented in OpenFOAM v1912 for testing and validation. 3D models of CATs were created using CT scans and computational meshes, and the results were compared to invasive coronary angiographic (ICA) data to validate the accuracy and effectiveness of the PBA. The discrepancy between the calculated and experimental FFR was within 2.33-5.26% in the steady-state and transient simulations, respectively, when convergence was reached. The PBA was a reliable and physiologically sound technique compared to a current lumped parameter model (LPM), which is based on empirical scaling correlations and requires nonlinear iterative computing for convergence. The accuracy of the PBA method was further confirmed using an FDA nozzle, which demonstrated good alignment with the CFD-validated values.

Myocardial Perfusion Single-Photon Emission Computed Tomography (SPECT) Image Denoising: A Comparative Study.

The present study aimed to evaluate the effectiveness of different filters in improving the quality of myocardial perfusion single-photon emission computed tomography (SPECT) images. Data were collected using the Siemens Symbia T2 dual-head SPECT/Computed tomography (CT) scanner. Our dataset included more than 900 images from 30 patients. The quality of the SPECT was evaluated after applying filters such as the Butterworth, Hamming, Gaussian, Wiener, and median-modified Wiener filters with different kernel sizes, by calculating indicators such as the signal-to-noise ratio (SNR), peak signal-to-noise ratio (PSNR), and contrast-to-noise ratio (CNR). SNR and CNR were highest with the Wiener filter with a kernel size of 5 × 5. Additionally, the Gaussian filter achieved the highest PSNR. The results revealed that the Wiener filter, with a kernel size of 5 × 5, outperformed the other filters for denoising images of our dataset. The novelty of this study includes comparison of different filters to improve the quality of myocardial perfusion SPECT. As far as we know, this is the first study to compare the mentioned filters on myocardial perfusion SPECT images, using our datasets with specific noise structures and mentioning all the elements necessary for its presentation within one document.

A computational framework for the multiphysics simulation of microbubble-mediated sonothrombolysis using a forward-viewing intravascular transducer.

Sonothrombolysis is a technique that utilises ultrasound waves to excite microbubbles surrounding a clot. Clot lysis is achieved through mechanical damage induced by acoustic cavitation and through local clot displacement induced by acoustic radiation force (ARF). Despite the potential of microbubble-mediated sonothrombolysis, the selection of the optimal ultrasound and microbubble parameters remains a challenge. Existing experimental studies are not able to provide a complete picture of how ultrasound and microbubble characteristics influence the outcome of sonothrombolysis. Likewise, computational studies have not been applied in detail in the context of sonothrombolysis. Hence, the effect of interaction between the bubble dynamics and acoustic propagation on the acoustic streaming and clot deformation remains unclear. In the present study, we report for the first time the computational framework that couples the bubble dynamic phenomena with the acoustic propagation in a bubbly medium to simulate microbubble-mediated sonothrombolysis using a forward-viewing transducer. The computational framework was used to investigate the effects of ultrasound properties (pressure and frequency) and microbubble characteristics (radius and concentration) on the outcome of sonothrombolysis. Four major findings were obtained from the simulation results: (i) ultrasound pressure plays the most dominant role over all the other parameters in affecting the bubble dynamics, acoustic attenuation, ARF, acoustic streaming, and clot displacement, (ii) smaller microbubbles could contribute to a more violent oscillation and improve the ARF simultaneously when they are stimulated at higher ultrasound pressure, (iii) higher microbubbles concentration increases the ARF, and (iv) the effect of ultrasound frequency on acoustic attenuation is dependent on the ultrasound pressure. These results may provide fundamental insight that is crucial in bringing sonothrombolysis closer to clinical implementation.

Comparing Different Algorithms for the Pseudo-Coloring of Myocardial Perfusion Single-Photon Emission Computed Tomography Images.

Single-photon emission computed tomography (SPECT) images can significantly help physicians in diagnosing patients with coronary artery or suspected coronary artery diseases. However, these images are grayscale with qualities that are not readily visible. The objective of this study was to evaluate the effectiveness of different pseudo-coloring algorithms of myocardial perfusion SPECT images. Data were collected using a Siemens Symbia T2 dual-head SPECT/computed tomography (CT) scanner. After pseudo-coloring, the images were assessed both qualitatively and quantitatively. The qualities of different pseudo-color images were examined by three experts, while the images were evaluated quantitatively by obtaining indices such as mean squared error (MSE), peak signal-to-noise ratio (PSNR), normalized color difference (NCD), and structure similarity index metric (SSIM). The qualitative evaluation demonstrated that the warm color map (WCM), followed by the jet color map, outperformed the remaining algorithms in terms of revealing the non-visible qualities of the images. Furthermore, the quantitative evaluation results demonstrated that the WCM had the highest PSNR and SSIM but the lowest MSE. Overall, the WCM could outperform the other color maps both qualitatively and quantitatively. The novelty of this study includes comparing different pseudo-coloring methods to improve the quality of myocardial perfusion SPECT images and utilizing our collected datasets.

Automated beat-wise arrhythmia diagnosis using modified U-net on extended electrocardiographic recordings with heterogeneous arrhythmia types.

Abnormality of the cardiac conduction system can induce arrhythmia - abnormal heart rhythm - that can frequently lead to other cardiac diseases and complications, and are sometimes life-threatening. These conduction system perturbations can manifest as morphological changes on the surface electrocardiographic (ECG) signal. Assessment of these morphological changes can be challenging and time-consuming, as ECG signal features are often low in amplitude and subtle. The main aim of this study is to develop an automated computer aided diagnostic (CAD) system that can expedite the process of arrhythmia diagnosis, as an aid to clinicians to provide appropriate and timely intervention to patients. We propose an autoencoder of ECG signals that can diagnose normal sinus beats, atrial premature beats (APB), premature ventricular contractions (PVC), left bundle branch block (LBBB) and right bundle branch block (RBBB). Apart from the first, the rest are morphological beat-to-beat elements that characterize and constitute complex arrhythmia. The novelty of this work lies in how we modified the U-net model to perform beat-wise analysis on heterogeneously segmented ECGs of variable lengths derived from the MIT-BIH arrhythmia database. The proposed system has demonstrated self-learning ability in generating class activations maps, and these generated maps faithfully reflect the cardiac conditions in each ECG cardiac cycle. It has attained a high classification accuracy of 97.32% in diagnosing cardiac conditions, and 99.3% for R peak detection using a ten-fold cross validation strategy. Our developed model can help physicians to screen ECG accurately, potentially resulting in timely intervention of patients with arrhythmia.

Automated diagnosis of arrhythmia using combination of CNN and LSTM techniques with variable length heart beats.

Arrhythmia is a cardiac conduction disorder characterized by irregular heartbeats. Abnormalities in the conduction system can manifest in the electrocardiographic (ECG) signal. However, it can be challenging and time-consuming to visually assess the ECG signals due to the very low amplitudes. Implementing an automated system in the clinical setting can potentially help expedite diagnosis of arrhythmia, and improve the accuracies. In this paper, we propose an automated system using a combination of convolutional neural network (CNN) and long short-term memory (LSTM) for diagnosis of normal sinus rhythm, left bundle branch block (LBBB), right bundle branch block (RBBB), atrial premature beats (APB) and premature ventricular contraction (PVC) on ECG signals. The novelty of this work is that we used ECG segments of variable length from the MIT-BIT arrhythmia physio bank database. The proposed system demonstrated high classification performance in the handling of variable-length data, achieving an accuracy of 98.10%, sensitivity of 97.50% and specificity of 98.70% using ten-fold cross validation strategy. Our proposed model can aid clinicians to detect common arrhythmias accurately on routine screening ECG.

Automated retinal health diagnosis using pyramid histogram of visual words and Fisher vector techniques.

Untreated age-related macular degeneration (AMD), diabetic retinopathy (DR), and glaucoma may lead to irreversible vision loss. Hence, it is essential to have regular eye screening to detect these eye diseases at an early stage and to offer treatment where appropriate. One of the simplest, non-invasive and cost-effective techniques to screen the eyes is by using fundus photo imaging. But, the manual evaluation of fundus images is tedious and challenging. Further, the diagnosis made by ophthalmologists may be subjective. Therefore, an objective and novel algorithm using the pyramid histogram of visual words (PHOW) and Fisher vectors is proposed for the classification of fundus images into their respective eye conditions (normal, AMD, DR, and glaucoma). The proposed algorithm extracts features which are represented as words. These features are built and encoded into a Fisher vector for classification using random forest classifier. This proposed algorithm is validated with both blindfold and ten-fold cross-validation techniques. An accuracy of 90.06% is achieved with the blindfold method, and highest accuracy of 96.79% is obtained with ten-fold cross-validation. The highest classification performance of our system shows the potential of deploying it in polyclinics to assist healthcare professionals in their initial diagnosis of the eye. Our developed system can reduce the workload of ophthalmologists significantly.

Computer aided diagnosis of diabetic foot using infrared thermography: A review.

Diabetes mellitus (DM) is a chronic metabolic disorder that requires regular medical care to prevent severe complications. The elevated blood glucose level affects the eyes, blood vessels, nerves, heart, and kidneys after the onset. The affected blood vessels (usually due to atherosclerosis) may lead to insufficient blood circulation particularly in the lower extremities and nerve damage (neuropathy), which can result in serious foot complications. Hence, an early detection and treatment can prevent foot complications such as ulcerations and amputations. Clinicians often assess the diabetic foot for sensory deficits with clinical tools, and the resulting foot severity is often manually evaluated. The infrared thermography is a fast, nonintrusive and non-contact method which allows the visualization of foot plantar temperature distribution. Several studies have proposed infrared thermography-based computer aided diagnosis (CAD) methods for diabetic foot. Among them, the asymmetric temperature analysis method is more superior, as it is easy to implement, and yielded satisfactory results in most of the studies. In this paper, the diabetic foot, its pathophysiology, conventional assessments methods, infrared thermography and the different infrared thermography-based CAD analysis methods are reviewed.

Quantitative means for differentiating renal obstruction by analysing renography by compartmental modelling of renal fluid flow rate.

OBJECTIVE: The aim of this study was to investigate the accuracy of using a newly developed index, the ratio of urine outflow to renal pelvis volume U/V2 (1/s), in evaluating renal obstruction and determining the severity of obstruction. PATIENTS AND METHODS: A total of 42 patients' renograms (80 kidneys) were studied. Compartmental modelling was used to model the behaviour of tracers flowing through the kidney. The derived model led to the formation of the normalized urine flow rate U/V2. An analysis was carried to test the accuracy of the developed index by comparing the developed model and the clinical evaluation of renograms. The Support Vector Machine algorithm was implemented to predict the renal obstruction status. RESULTS: From the comparison performed between the index and the clinical evaluation from certified experts, it was shown that a higher value of index U/V2 indicated a normal kidney, whereas a lower value indicated an obstructed kidney. The classifier developed could provide a 100% accurate diagnosis of differentiated unobstructed kidneys (42/42) and obstructed kidney (18/18). For further classification of obstructed kidneys, the system grouped the samples into slightly obstructed cases with an accuracy of 100% (9/9) and heavily obstructed cases with an accuracy of 89% (8/9). CONCLUSION: The use of the single parameter U/V2 could produce the diagnosis of renal obstruction with a high level of accuracy. This method has the potential to be used as a benchmark to distinguish the severity level of the renal obstruction.

Preliminary investigation on the reduction of plantar loading pressure with different insole materials (SRP--Slow Recovery Poron, P--Poron, PPF--Poron +Plastazote, firm and PPS--Poron+Plastazote, soft).

BACKGROUND: The aim of this study is to investigate the amount of pressure reduction for different padding and insole materials commonly used in the podiatry clinic. METHODS: Plantar pressure were taken for 5 subjects without insoles fitted (BF) in their daily sports shoes, and thereafter with 4 pairs of simple insoles (6.4 mm thick) each as follow: SRP - Slow Recovery Poron, P - Poron, PPF - Poron+Plastazote (firm) and PPS - Poron+Plastazote (soft). In addition, subjects were also tested with semi-compressed felt (SCF) padding with a 1st metatarsophalangeal joint (MTPJ) aperture cut-out bilaterally. Minimum, maximum, mean pressure and peak pressure at the hallux, 1st, 2nd, 3rd/4th and 5th MTPJ across both feet were analysed. Repeated measures ANOVA with post hoc Bonferroni paired wise comparison was used to test for any statistical significance at the 95% confidence level for all pressure data. RESULTS: PPF was significant in reducing the minimum (p<0.005) and mean pressure (p<0.03) when compared to BF. This accounted for approximately 28% and 27% pressure reduction in minimum and mean pressure respectively. Peak pressure on the 1st MTPJ locality showed significant reduction of 37% and 29% with the use of SCF (p<0.004) and PPF (p<0.004), respectively. CONCLUSIONS: All 4 commonly used insole materials were able to reduce pressure across the whole foot with PPF achieving significance. Off-loading the 1st MTPJ would still be best achieved with the commonly used plantar metatarsal pad of SCF with the aperture cut-out design.

Explaining left ventricular pressure dynamics in terms of LV passive and active elastances.

There has been much characterization of the heart as a pump by means of models based on elastance and compliance. The present paper puts forward the new concept of time-varying passive and active elastance. The biomechanical basis of cyclic elastances of the left ventricle (LV) is presented. Elastance is defined in terms of the relationship between ventricular pressure and volume as dP = EdV+ VdE, where E includes passive elastance, Ep, and active elastance, Ea. By incorporating this concept in LV models to simulate diastolic (filling) and systolic phases, a time-varying expression has been obtained for Ea, and an LV volume dependent expression has been obtained for Ep. It is proposed to use these two elastances Ea and Ep to represent the intrinsic LV properties. The active elastance, Ea, can be used to characterize the LV contractile state and represents LV pressure variation due to LV volume variation (such as during the filling and ejection phases). The passive elastance, Ep, can serve as a measure of LV resistance to filling. Furthermore, it has been demonstrated how the LV pressure dynamics (and LV pressure response to LV volume) can be explained in terms of Ea and Ep.

Parametric study of the biopotential equation for breast tumour identification using ANOVA and Taguchi method.

Extensive literatures have shown significant trend of progressive electrical changes according to the proliferative characteristics of breast epithelial cells. Physiologists also further postulated that malignant transformation resulted from sustained depolarization and a failure of the cell to repolarize after cell division, making the area where cancer develops relatively depolarized when compared to their non-dividing or resting counterparts. In this paper, we present a new approach, the Biofield Diagnostic System (BDS), which might have the potential to augment the process of diagnosing breast cancer. This technique was based on the efficacy of analysing skin surface electrical potentials for the differential diagnosis of breast abnormalities. We developed a female breast model, which was close to the actual, by considering the breast as a hemisphere in supine condition with various layers of unequal thickness. Isotropic homogeneous conductivity was assigned to each of these compartments and the volume conductor problem was solved using finite element method to determine the potential distribution developed due to a dipole source. Furthermore, four important parameters were identified and analysis of variance (ANOVA, Yates' method) was performed using design (n = number of parameters, 4). The effect and importance of these parameters were analysed. The Taguchi method was further used to optimise the parameters in order to ensure that the signal from the tumour is maximum as compared to the noise from other factors. The Taguchi method used proved that probes' source strength, tumour size and location of tumours have great effect on the surface potential field. For best results on the breast surface, while having the biggest possible tumour size, low amplitudes of current should be applied nearest to the breast surface.

Left ventricular shape-based contractility index.

This study develops contractility indices in terms of the left ventricular (LV) ellipsoidal geometrical shape-factor. The contractility index (CONT1) is given by the maximum value dsigma(*)/dt wherein sigma(*)=sigma/P, sigma is the wall stress, and sigma(*) is expressed in terms of the shape factor S (the ratio of the minor axis and major axis, B/A, of the instantaneous LV ellipsoidal model). Another contractility index (CONT2) is also developed based on how far apart the in vivo S at the start of ejection is from its optimized value, CONT2=(S(se)-S(se)(op))/S(se)(op), where S(se) refers to the value of S at the start of ejection, S(se)(op) is the derived optimal value of S(se) for which sigma* is maximum. The values of S(=B/A) were calculated from cineventriculographically monitored LV volume, myocardial volume and wall-thickness. Then both the contractility indices were evaluated in normal subjects, as well as in patients with mild heart failure and in patients with severe heart failure. The normal values of CONT1 and CONT2 are 8.75+/-2.30s(-1) and 0.09+/-0.07, respectively. CONT1 decreased in patients with mild and severe heart failures to 5.78+/-1.30 and 3.90+/-1.30, respectively. CONT2 increased in patients with mild and severe heart failures to 0.11+/-0.09 and 0.23+/-0.12, respectively. This implies that a non-optimal and less ellipsoidal shape is associated with decreased contractility (and poor systolic function) of the LV. CONT1 and CONT2 are useful as non-invasively determinable quantitative indices of LV contractility, to distinguish between normal and pathologic LVs.

Passive and active ventricular elastances of the left ventricle.

BACKGROUND: Description of the heart as a pump has been dominated by models based on elastance and compliance. Here, we are presenting a somewhat new concept of time-varying passive and active elastance. The mathematical basis of time-varying elastance of the ventricle is presented. We have defined elastance in terms of the relationship between ventricular pressure and volume, as: dP = EdV + VdE, where E includes passive (Ep) and active (Ea) elastance. By incorporating this concept in left ventricular (LV) models to simulate filling and systolic phases, we have obtained the time-varying expression for Ea and the LV-volume dependent expression for Ep. METHODS AND RESULTS: Using the patient's catheterization-ventriculogram data, the values of passive and active elastance are computed. Ea is expressed as [formula: see text] Epis represented as: [formula: see text]. Ea is deemed to represent a measure of LV contractility. Hence, Peak dP/dt and ejection fraction (EF) are computed from the monitored data and used as the traditional measures of LV contractility. When our computed peak active elastance (Ea,max) is compared against these traditional indices by linear regression, a high degree of correlation is obtained. As regards Ep, it constitutes a volume-dependent stiffness property of the LV, and is deemed to represent resistance-to-filling. CONCLUSIONS: Passive and active ventricular elastance formulae can be evaluated from a single-beat P-V data by means of a simple-to-apply LV model. The active elastance (Ea) can be used to characterize the ventricle's contractile state, while passive elastance (Ep) can represent a measure of resistance-to-filling.