An XAI approach for COVID-19 detection using transfer learning with X-ray images.

The coronavirus disease (COVID-19) has continued to cause severe challenges during this unprecedented time, affecting every part of daily life in terms of health, economics, and social development. There is an increasing demand for chest X-ray (CXR) scans, as pneumonia is the primary and vital complication of COVID-19. CXR is widely used as a screening tool for lung-related diseases due to its simple and relatively inexpensive application. However, these scans require expert radiologists to interpret the results for clinical decisions, i.e., diagnosis, treatment, and prognosis. The digitalization of various sectors, including healthcare, has accelerated during the pandemic, with the use and importance of Artificial Intelligence (AI) dramatically increasing. This paper proposes a model using an Explainable Artificial Intelligence (XAI) technique to detect and interpret COVID-19 positive CXR images. We further analyze the impact of COVID-19 positive CXR images using heatmaps. The proposed model leverages transfer learning and data augmentation techniques for faster and more adequate model training. Lung segmentation is applied to enhance the model performance further. We conducted a pre-trained network comparison with the highest classification performance (F1-Score: 98%) using the ResNet model.

Simultaneous Segmentation and Classification of Pressure Injury Image Data Using Mask-R-CNN.

Background: Pressure injuries (PIs) impose a substantial burden on patients, caregivers, and healthcare systems, affecting an estimated 3 million Americans and costing nearly $18 billion annually. Accurate pressure injury staging remains clinically challenging. Over the last decade, object detection and semantic segmentation have evolved quickly with new methods invented and new application areas emerging. Simultaneous object detection and segmentation paved the way to segment and classify anatomical structures. In this study, we utilize the Mask-R-CNN algorithm for segmentation and classification of stage 1-4 pressure injuries. Methods: Images from the eKare Inc. pressure injury wound data repository were segmented and classified manually by two study authors with medical training. The Mask-R-CNN model was implemented using the Keras deep learning and TensorFlow libraries with Python. We split 969 pressure injury images into training (87.5%) and validation (12.5%) subsets for Mask-R-CNN training. Results: We included 121 random pressure injury images in our test set. The Mask-R-CNN model showed overall classification accuracy of 92.6%, and the segmentation demonstrated 93.0% accuracy. Our F1 scores for stages 1-4 were 0.842, 0.947, 0.907, and 0.944, respectively. Our Dice coefficients for stages 1-4 were 0.92, 0.85, 0.93, and 0.91, respectively. Conclusions: Our Mask-R-CNN model provides levels of accuracy considerably greater than the average healthcare professional who works with pressure injury patients. This tool can be easily incorporated into the clinician's workflow to aid in the hospital setting.

Enhancing Patient Centricity and Advancing Innovation in Clinical Research with Virtual Randomized Clinical Trials (vRCTs).

The current public health crisis has highlighted the need to accelerate healthcare innovation. Despite unwavering levels of cooperation among academia, industry, and policy makers, it can still take years to bring a life-saving product to market. There are some obvious limitations, including lack of blinding or masking and small sample size, which render the results less applicable to the real world. Traditional randomized controlled trials (RCTs) are lengthy, expensive, and have a low success rate. There is a growing acknowledgement that the current process no longer fully meets the growing healthcare needs. Advances in technology coupled with proliferation of telehealth modalities, sensors, wearable and connected devices have paved the way for a new paradigm. Virtual randomized controlled trials (vRCTs) have the potential to drastically shorten the clinical trial cycle while maximizing patient-centricity, compliance, and recruitment. This new approach can inform clinical trials in real time and with a holistic view of a patient's health. This paper provides an overview of virtual clinical trials, addressing critical issues, including regulatory compliance, data security, privacy, and ownership.

Radiation-free methods for navigated screw placement in slipped capital femoral epiphysis surgery.

PURPOSE: For orthopedic procedures, surgeons utilize intra-operative medical images such as fluoroscopy to plan screw placement and accurately position the guide wire with the intended trajectory. The number of fluoroscopic images needed depends on the complexity of the case and skill of the surgeon. Since more fluoroscopic images lead to more exposure and higher radiation dose for both surgeon and patient, a solution that decreases the number of fluoroscopic images would be an improvement in clinical care. METHODS: This article describes and compares three different novel navigation methods and techniques for screw placement using an attachable Inertial Measurement Unit device or a robotic arm. These methods provide projection and visualization of the surgical tool trajectory during the slipped capital femoral epiphysis procedure. RESULTS: These techniques resulted in faster and more efficient preoperative calibration and set up times compared to other intra-operative navigation systems in our phantom study. We conducted an experiment using 120 model bones to measure the accuracy of the methods. CONCLUSION: As conclusion, these approaches have the potential to improve accuracy of surgical tool navigation and decrease the number of required X-ray images without any change in the clinical workflow. The results also show 65% decrease in total error compared to the conventional manual approach.

Improved screw placement for slipped capital femoral epiphysis (SCFE) using robotically-assisted drill guidance.

Slipped Capital Femoral Epiphysis (SCFE) is a common hip displacement condition in adolescents. In the standard treatment, the surgeon uses intra-operative fluoroscopic imaging to plan the screw placement and the drill trajectory. The accuracy, duration, and efficacy of this procedure are highly dependent on surgeon skill. Longer procedure times result in higher radiation dose, to both patient and surgeon. A robotic system to guide the drill trajectory might help to reduce screw placement errors and procedure time by reducing the number of passes and confirmatory fluoroscopic images needed to verify accurate positioning of the drill guide along a planned trajectory. Therefore, with the long-term goals of improving screw placement accuracy, reducing procedure time and intra-operative radiation dose, our group is developing an image-guided robotic surgical system to assist a surgeon with pre-operative path planning and intra-operative drill guide placement.

Interactive initialization of 2D/3D rigid registration.

PURPOSE: Registration is one of the key technical components in an image-guided navigation system. A large number of 2D/3D registration algorithms have been previously proposed, but have not been able to transition into clinical practice. The authors identify the primary reason for the lack of adoption with the prerequisite for a sufficiently accurate initial transformation, mean target registration error of about 10 mm or less. In this paper, the authors present two interactive initialization approaches that provide the desired accuracy for x-ray/MR and x-ray/CT registration in the operating room setting. METHODS: The authors have developed two interactive registration methods based on visual alignment of a preoperative image, MR, or CT to intraoperative x-rays. In the first approach, the operator uses a gesture based interface to align a volume rendering of the preoperative image to multiple x-rays. The second approach uses a tracked tool available as part of a navigation system. Preoperatively, a virtual replica of the tool is positioned next to the anatomical structures visible in the volumetric data. Intraoperatively, the physical tool is positioned in a similar manner and subsequently used to align a volume rendering to the x-ray images using an augmented reality (AR) approach. Both methods were assessed using three publicly available reference data sets for 2D/3D registration evaluation. RESULTS: In the authors' experiments, the authors show that for x-ray/MR registration, the gesture based method resulted in a mean target registration error (mTRE) of 9.3 ± 5.0 mm with an average interaction time of 146.3 ± 73.0 s, and the AR-based method had mTREs of 7.2 ± 3.2 mm with interaction times of 44 ± 32 s. For x-ray/CT registration, the gesture based method resulted in a mTRE of 7.4 ± 5.0 mm with an average interaction time of 132.1 ± 66.4 s, and the AR-based method had mTREs of 8.3 ± 5.0 mm with interaction times of 58 ± 52 s. CONCLUSIONS: Based on the authors' evaluation, the authors conclude that the registration approaches are sufficiently accurate for initializing 2D/3D registration in the OR setting, both when a tracking system is not in use (gesture based approach), and when a tracking system is already in use (AR based approach).

Image-guided navigation: a cost effective practical introduction using the Image-Guided Surgery Toolkit (IGSTK).

Teaching the key technical aspects of image-guided interventions using a hands-on approach is a challenging task. This is primarily due to the high cost and lack of accessibility to imaging and tracking systems. We provide a software and data infrastructure which addresses both challenges. Our infrastructure allows students, patients, and clinicians to develop an understanding of the key technologies by using them, and possibly by developing additional components and integrating them into a simple navigation system which we provide. Our approach requires minimal hardware, LEGO blocks to construct a phantom for which we provide CT scans, and a webcam which when combined with our software provides the functionality of a tracking system. A premise of this approach is that tracking accuracy is sufficient for our purpose. We evaluate the accuracy provided by a consumer grade webcam and show that it is sufficient for educational use. We provide an open source implementation of all the components required for a basic image-guided navigation as part of the Image-Guided Surgery Toolkit (IGSTK). It has long been known that in education there is no substitute for hands-on experience, to quote Sophocles, "One must learn by doing the thing; for though you think you know it, you have no certainty, until you try.". Our work provides this missing capability in the context of image-guided navigation. Enabling a wide audience to learn and experience the use of a navigation system.

ADC histograms predict response to anti-angiogenic therapy in patients with recurrent high-grade glioma.

INTRODUCTION: The purpose of this study is to evaluate apparent diffusion coefficient (ADC) maps to distinguish anti-vascular and anti-tumor effects in the course of anti-angiogenic treatment of recurrent high-grade gliomas (rHGG) as compared to standard magnetic resonance imaging (MRI). METHODS: This retrospective study analyzed ADC maps from diffusion-weighted MRI in 14 rHGG patients during bevacizumab/irinotecan (B/I) therapy. Applying image segmentation, volumes of contrast-enhanced lesions in T1 sequences and of hyperintense T2 lesions (hT2) were calculated. hT2 were defined as regions of interest (ROI) and registered to corresponding ADC maps (hT2-ADC). Histograms were calculated from hT2-ADC ROIs. Thereafter, histogram asymmetry termed "skewness" was calculated and compared to progression-free survival (PFS) as defined by the Response Assessment Neuro-Oncology (RANO) Working Group criteria. RESULTS: At 8-12 weeks follow-up, seven (50%) patients showed a partial response, three (21.4%) patients were stable, and four (28.6%) patients progressed according to RANO criteria. hT2-ADC histograms demonstrated statistically significant changes in skewness in relation to PFS at 6 months. Patients with increasing skewness (n = 11) following B/I therapy had significantly shorter PFS than did patients with decreasing or stable skewness values (n = 3, median percentage change in skewness 54% versus -3%, p = 0.04). CONCLUSION: In rHGG patients, the change in ADC histogram skewness may be predictive for treatment response early in the course of anti-angiogenic therapy and more sensitive than treatment assessment based solely on RANO criteria.