Harnessing Speech-Derived Digital Biomarkers to Detect and Quantify Cognitive Decline Severity in Older Adults.

INTRODUCTION: Current cognitive assessments suffer from floor/ceiling and practice effects, poor psychometric performance in mild cases, and repeated assessment effects. This study explores the use of digital speech analysis as an alternative tool for determining cognitive impairment. The study specifically focuses on identifying the digital speech biomarkers associated with cognitive impairment and its severity. METHODS: We recruited older adults with varying cognitive health. Their speech data, recorded via a wearable microphone during the reading aloud of a standard passage, were processed to derive digital biomarkers such as timing, pitch, and loudness. Cohen's d effect size highlighted group differences, and correlations were drawn to the Montreal Cognitive Assessment (MoCA). A stepwise approach using a Random Forest model was implemented to distinguish cognitive states using speech data and predict MoCA scores based on highly correlated features. RESULTS: The study comprised 59 participants, with 36 demonstrating cognitive impairment and 23 serving as cognitively intact controls. Among all assessed parameters, similarity, as determined by Dynamic Time Warping (DTW), exhibited the most substantial positive correlation (rho = 0.529, p < 0.001), while timing parameters, specifically the ratio of extra words, revealed the strongest negative correlation (rho = -0.441, p < 0.001) with MoCA scores. Optimal discriminative performance was achieved with a combination of four speech parameters: total pause time, speech-to-pause ratio, similarity via DTW, and intelligibility via DTW. Precision and balanced accuracy scores were found to be 88.1 ± 1.2% and 76.3 ± 1.3%, respectively. DISCUSSION: Our research proposes that reading-derived speech data facilitates the differentiation between cognitively impaired individuals and cognitively intact, age-matched older adults. Specifically, parameters based on timing and similarity within speech data provide an effective gauge of cognitive impairment severity. These results suggest speech analysis as a viable digital biomarker for early detection and monitoring of cognitive impairment, offering novel approaches in dementia care.

Remote at-home wearable-based gait assessments in Progressive Supranuclear Palsy compared to Parkinson's Disease.

BACKGROUND: Wearable sensors can differentiate Progressive Supranuclear Palsy (PSP) from Parkinson's Disease (PD) in laboratory settings but have not been tested in remote settings. OBJECTIVES: To compare gait and balance in PSP and PD remotely using wearable-based assessments. METHODS: Participants with probable PSP or probable/clinically established PD with reliable caregivers, still able to ambulate 10 feet unassisted, were recruited, enrolled, and consented remotely and instructed by video conference to operate a study-specific tablet solution (BioDigit Home ™) and to wear three inertial sensors (LEGSys™, BioSensics LLC, Newton, MA USA) while performing the Timed Up and Go, 5 × sit-to-stand, and 2-min walk tests. PSPRS and MDS-UPDRS scores were collected virtually or during routine clinical visits. RESULTS: Between November, 2021- November, 2022, 27 participants were screened of whom 3 were excluded because of technological difficulties. Eleven PSP and 12 PD participants enrolled, of whom 10 from each group had complete analyzable data. Demographics were well-matched (PSP mean age = 67.6 ± 1.3 years, 40% female; PD mean age = 70.3 ± 1.8 years, 40% female) while disease duration was significantly shorter in PSP (PSP 14 ± 3.5 months vs PD 87.9 ± 16.9 months). Gait parameters showed significant group differences with effect sizes ranging from d = 1.0 to 2.27. Gait speed was significantly slower in PSP: 0.45 ± 0.06 m/s vs. 0.79 ± 0.06 m/s in PD (d = 1.78, p < 0.001). CONCLUSION: Our study demonstrates the feasibility of measuring gait in PSP and PD remotely using wearable sensors. The study provides insight into digital biomarkers for both neurodegenerative diseases. TRIAL REGISTRATION: ClinicalTrials.gov Identifier: NCT04753320, first posted Febuary 15, 2021.

Dual-Task Upper Extremity Motor Performance Measured by Video Processing as Cognitive-Motor Markers for Older Adults.

INTRODUCTION: The use of dual-task model such as dual-task gait has been extensively studied to assess cognitive-motor performance among older adults. However, space restriction and safety factor limit its applications in remote assessment. To address the gap, we propose a video processing-based approach to remotely quantify cognitive-motor performance using a 20-s repetitive elbow flexion-extension test with dual-task condition, called video-based motoric-cognitive meter (MCM). METHODS: Eighteen older participants (age: 78.6 ± 6.5 years) who were clinically diagnosed as having either mild cognitive impairment or dementia were included in this study. Participants were asked to perform 20-s repetitive elbow flexion-extension exercise with a memory exercise by counting backward from a two-digit number. During the test, all movements of the forearm were recorded by a video camera. As a comparator, a validated wrist-worn sensor was used, which allowed quantifying upper extremity kinematics. RESULTS: The results showed a good agreement (r ≥ 0.530 and ICC2,1 ≥ 0.681) between the derived dual-task upper extremity motor performance from the proposed video-based MCM and a clinically validated sensor-based MCM. We also observed moderate correlations (r ≥ 0.496) between some measures of video-based MCM (flexion time, extension time, and flexion-extension time) and clinical cognitive scale (Mini-Mental State Examination [MMSE]). Additionally, some measures of dual-task upper extremity motor performance (speed, flexion time, extension time, and flexion-extension time) were associated with dual-task gait speed (r ≥ 0.557), which has been found to be correlated with cognitive impairment. Lastly, the selected dual-task motor performance metric (flexion time) was sensitive to predict MMSE scores in linear regression analyses with statistical significance (adjusted R2 = 0.306, p = 0.025). CONCLUSION: This study proposes a video processing-based approach to analyze dual-task upper extremity motor performance from a simple and convenient upper extremity function test. The results indicate concurrent validity of the proposed video-based MCM compared with the sensor-based MCM, and associations between dual-task upper extremity motor performance and clinically validated cognitive markers (MMSE scores and dual-task gait). Future studies are warranted to explore sensitivity of this solution to promote remote assessment of cognitive-motor performance among older adults in telehealth applications.

Automated Bone Tumor Segmentation and Classification as Benign or Malignant Using Computed Tomographic Imaging.

The purpose of this study was to pair computed tomography (CT) imaging and machine learning for automated bone tumor segmentation and classification to aid clinicians in determining the need for biopsy. In this retrospective study (March 2005-October 2020), a dataset of 84 femur CT scans (50 females and 34 males, 20 years and older) with definitive histologic confirmation of bone lesion (71% malignant) were leveraged to perform automated tumor segmentation and classification. Our method involves a deep learning architecture that receives a DICOM slice and predicts (i) a segmentation mask over the estimated tumor region, and (ii) a corresponding class as benign or malignant. Class prediction for each case is then determined via majority voting. Statistical analysis was conducted via fivefold cross validation, with results reported as averages along with 95% confidence intervals. Despite the imbalance between benign and malignant cases in our dataset, our approach attains similar classification performances in specificity (75%) and sensitivity (79%). Average segmentation performance attains 56% Dice score and reaches up to 80% for an image slice in each scan. The proposed approach establishes the first steps in developing an automated deep learning method on bone tumor segmentation and classification from CT imaging. Our approach attains comparable quantitative performance to existing deep learning models using other imaging modalities, including X-ray. Moreover, visual analysis of bone tumor segmentation indicates that our model is capable of learning typical tumor characteristics and provides a promising direction in aiding the clinical decision process for biopsy.

Using Deep Learning to Detect the Presence and Location of Hemoperitoneum on the Focused Assessment with Sonography in Trauma (FAST) Examination in Adults.

Abdominal ultrasonography has become an integral component of the evaluation of trauma patients. Internal hemorrhage can be rapidly diagnosed by finding free fluid with point-of-care ultrasound (POCUS) and expedite decisions to perform lifesaving interventions. However, the widespread clinical application of ultrasound is limited by the expertise required for image interpretation. This study aimed to develop a deep learning algorithm to identify the presence and location of hemoperitoneum on POCUS to assist novice clinicians in accurate interpretation of the Focused Assessment with Sonography in Trauma (FAST) exam. We analyzed right upper quadrant (RUQ) FAST exams obtained from 94 adult patients (44 confirmed hemoperitoneum) using the YoloV3 object detection algorithm. Exams were partitioned via fivefold stratified sampling for training, validation, and hold-out testing. We assessed each exam image-by-image using YoloV3 and determined hemoperitoneum presence for the exam using the detection with highest confidence score. We determined the detection threshold as the score that maximizes the geometric mean of sensitivity and specificity over the validation set. The algorithm had 95% sensitivity, 94% specificity, 95% accuracy, and 97% AUC over the test set, significantly outperforming three recent methods. The algorithm also exhibited strength in localization, while the detected box sizes varied with a 56% IOU averaged over positive cases. Image processing demonstrated only 57-ms latency, which is adequate for real-time use at the bedside. These results suggest that a deep learning algorithm can rapidly and accurately identify the presence and location of free fluid in the RUQ of the FAST exam in adult patients with hemoperitoneum.

Assessment of gait and balance impairment in people with spinocerebellar ataxia using wearable sensors.

OBJECTIVE: To explore the use of wearable sensors for objective measurement of motor impairment in spinocerebellar ataxia (SCA) patients during clinical assessments of gait and balance. METHODS: In total, 14 patients with genetically confirmed SCA (mean age 61.6 ± 8.6 years) and 4 healthy controls (mean age 49.0 ± 16.4 years) were recruited through the Massachusetts General Hospital (MGH) Ataxia Center. Participants donned seven inertial sensors while performing two independent trials of gait and balance assessments from the Scale for the Assessment and Rating of Ataxia (SARA) and Brief Ataxia Rating Scale (BARS2). Univariate analysis was used to identify sensor-derived metrics from wearable sensors that discriminate motor function between the SCA and control groups. Multivariate linear regression models were used to estimate the subjective in-person SARA/BARS2 ratings. Spearman correlation coefficients were used to evaluate the performance of the model. RESULTS: Stride length variability, stride duration, cadence, stance phase, pelvis sway, and turn duration were different between SCA and controls (p < 0.05). Similarly, sway and sway velocity of the ankle, hip, and center of mass differentiated SCA and controls (p < 0.05). Using these features, linear regression models showed moderate-to-strong correlation with clinical scores from the in-person rater during SARA assessments of gait (r = 0.73, p = 0.003) and stance (r = 0.90, p < 0.001) and the BARS2 gait assessment (r = 0.74, p = 0.003). CONCLUSION: This study demonstrates that sensor-derived metrics can potentially be used to estimate the level of motor impairment in patient with SCA quickly and objectively. Thus, digital biomarkers from wearable sensors have the potential to be an integral tool for SCA clinical trials and care.

Objective Assessment of Upper-Extremity Motor Functions in Spinocerebellar Ataxia Using Wearable Sensors.

The study presents a novel approach to objectively assessing the upper-extremity motor symptoms in spinocerebellar ataxia (SCA) using data collected via a wearable sensor worn on the patient's wrist during upper-extremity tasks associated with the Assessment and Rating of Ataxia (SARA). First, we developed an algorithm for detecting/extracting the cycles of the finger-to-nose test (FNT). We extracted multiple features from the detected cycles and identified features and parameters correlated with the SARA scores. Additionally, we developed models to predict the severity of symptoms based on the FNT. The proposed technique was validated on a dataset comprising the seventeen (n = 17) participants' assessments. The cycle detection technique showed an accuracy of 97.6% in a Bland-Altman analysis and a 94% accuracy (F1-score of 0.93) in predicting the severity of the FNT. Furthermore, the dependency of the upper-extremity tests was investigated through statistical analysis, and the results confirm dependency and potential redundancies in the upper-extremity SARA assessments. Our findings pave the way to enhance the utility of objective measures of SCA assessments. The proposed wearable-based platform has the potential to eliminate subjectivity and inter-rater variabilities in assessing ataxia.

Digital Biomarkers of Gait and Balance in Diabetic Foot, Measurable by Wearable Inertial Measurement Units: A Mini Review.

People with diabetic foot frequently exhibit gait and balance dysfunction. Recent advances in wearable inertial measurement units (IMUs) enable to assess some of the gait and balance dysfunction associated with diabetic foot (i.e., digital biomarkers of gait and balance). However, there is no review to inform digital biomarkers of gait and balance dysfunction related to diabetic foot, measurable by wearable IMUs (e.g., what gait and balance parameters can wearable IMUs collect? Are the measurements repeatable?). Accordingly, we conducted a web-based, mini review using PubMed. Our search was limited to human subjects and English-written papers published in peer-reviewed journals. We identified 20 papers in this mini review. We found preliminary evidence of digital biomarkers of gait and balance dysfunction in people with diabetic foot, such as slow gait speed, large gait variability, unstable gait initiation, and large body sway. However, due to heterogeneities in included papers in terms of study design, movement tasks, and small sample size, more studies are recommended to confirm this preliminary evidence. Additionally, based on our mini review, we recommend establishing appropriate strategies to successfully incorporate wearable-based assessment into clinical practice for diabetic foot care.

Smart-Home Concept for Remote Monitoring of Instrumental Activities of Daily Living (IADL) in Older Adults with Cognitive Impairment: A Proof of Concept and Feasibility Study.

Assessment of instrumental activities of daily living (IADL) is essential for the diagnosis and staging of dementia. However, current IADL assessments are subjective and cannot be administered remotely. We proposed a smart-home design, called IADLSys, for remote monitoring of IADL. IADLSys consists of three major components: (1) wireless physical tags (pTAG) attached to objects of interest, (2) a pendant-sensor to monitor physical activities and detect interaction with pTAGs, and (3) an interactive tablet as a gateway to transfer data to a secured cloud. Four studies, including an exploratory clinical study with five older adults with clinically confirmed cognitive impairment, who used IADLSys for 24 h/7 days, were performed to confirm IADLSys feasibility, acceptability, adherence, and validity of detecting IADLs of interest and physical activity. Exploratory tests in two cases with severe and mild cognitive impairment, respectively, revealed that a case with severe cognitive impairment either overestimated or underestimated the frequency of performed IADLs, whereas self-reporting and objective IADL were comparable for the case with mild cognitive impairment. This feasibility and acceptability study may pave the way to implement the smart-home concept to remotely monitor IADL, which in turn may assist in providing personalized support to people with cognitive impairment, while tracking the decline in both physical and cognitive function.

Mobility Performance in Community-Dwelling Older Adults: Potential Digital Biomarkers of Concern about Falling.

INTRODUCTION: Concern about falling is a prevalent worry among community-dwelling older adults and may contribute to a decline in physical and mental health. This study aimed to examine the association between mobility performance and concern about falling. METHODS: Older adults aged 65 years and older, with Mini-Mental State Examination score ≥24, and ambulatory (with or without the assistive device) were included. Concern about falling was evaluated with Falls Efficacy Scale-International (FES-I) scores. Participants with high concern about falling were identified using the cutoff of FES-I ≥23. Participants' motor capacity was assessed in standardized walking tests under single- and dual-task conditions. Participants' mobility performance was measured based on a 48-h trunk accelerometry signal from a wearable pendant sensor. RESULTS: No significant differences were observed at participant characteristics across groups with different levels of concern about falling (low: N = 64, age = 76.3 ± 7.2 years, female = 46%; high: N = 59, age = 79.3 ± 9.1 years, female = 47%), after propensity matching with BMI, age, depression, and cognition. With adjustment of motor capacity (stride velocity and stride length under single- and dual-task walking conditions), participants with high concern about falling had significantly poorer mobility performance than those with low concern about falling, including lower walking quantity (walking bouts, steps and time per day, and walking bout average, walking bout variability, and longest walking bout, p ≤ 0.013), and poorer daily-life gait (stride velocity and gait variability, p ≤ 0.023), and poorer walking quality (frontal gait symmetry, and trunk acceleration and velocity intensity, p ≤ 0.041). The selected mobility performance metrics (daily steps and frontal gait symmetry) could significantly contribute to identifying older adults with high concern about falling (p ≤ 0.042), having better model performance (p = 0.036) than only walking quantity (daily steps) with adjustment of confounding effects from the motor capacity (stride length under dual-task walking condition). CONCLUSION: There is an association between mobility performance and concern about falling in older adults. Mobility performance metrics can serve as predictors to identify older adults with high concern about falling, potentially providing digital biomarkers for clinicians to remotely track older adults' change of concern about falling via applications of remote patient monitoring.

Reprogrammable Braille on an elastic shell.

We describe a minimal realization of reversibly programmable matter in the form of a featureless smooth elastic plate that has the capacity to store information in a Braille-like format as a sequence of stable discrete dimples. Simple experiments with cylindrical and spherical shells show that we can control the number, location, and the temporal order of these dimples, which can be written and erased at will. Theoretical analysis of the governing equations in a specialized setting and numerical simulations of the complete equations allow us to characterize the phase diagram for the formation of these localized elastic states, elastic bits (e-bits), consistent with our observations. Given that the inherent bistability and hysteresis in these low-dimensional systems arise exclusively due to the geometrical-scale separation, independent of material properties or absolute scale, our results might serve as alternate approaches to small-scale mechanical memories.

Computational modeling of human bone fracture healing affected by different conditions of initial healing stage.

BACKGROUND: Bone healing process includes four phases: inflammatory response, soft callus formation, hard callus development, and remodeling. Mechanobiological models have been used to investigate the role of various mechanical and biological factors on bone healing. However, the effects of initial healing phase, which includes the inflammatory stage, the granulation tissue formation, and the initial callus formation during the first few days post-fracture, are generally neglected in such studies. METHODS: In this study, we developed a finite-element-based model to simulate different levels of diffusion coefficient for mesenchymal stem cell (MSC) migration, Young's modulus of granulation tissue, callus thickness and interfragmentary gap size to understand the modulatory effects of these initial phase parameters on bone healing. RESULTS: The results quantified how faster MSC migration, stiffer granulation tissue, thicker callus, and smaller interfragmentary gap enhanced healing to some extent. However, after a certain threshold, a state of saturation was reached for MSC migration rate, granulation tissue stiffness, and callus thickness. Therefore, a parametric study was performed to verify that the callus formed at the initial phase, in agreement with experimental observations, has an ideal range of geometry and material properties to have the most efficient healing time. CONCLUSIONS: Findings from this paper quantified the effects of the initial healing phase on healing outcome to better understand the biological and mechanobiological mechanisms and their utilization in the design and optimization of treatment strategies. It is also demonstrated through a simulation that for fractures, where bone segments are in close proximity, callus development is not required. This finding is consistent with the concepts of primary and secondary bone healing.

Influence of disruption of the acromioclavicular and coracoclavicular ligaments on glenohumeral motion: a kinematic evaluation.

BACKGROUND: Changes to the integrity of the acromioclavicular (AC) joint impact scapulothoracic and clavicular kinematics. AC ligaments provide anterior-posterior stability, while the coracoclavicular (CC) ligaments provide superior-inferior stability and a restraint to scapular internal rotation. The purpose of this cadaveric study was to describe the effect of sequential AC and CC sectioning on glenohumeral (GH) kinematics during abduction (ABD) of the arm. We hypothesized that complete AC ligament insult would result in altered GH translation in the anterior-posterior plane during abduction, while subsequent sectioning of both CC ligaments would result in an increasing inferior shift in GH translation. METHODS: Six cadaveric shoulders were studied to evaluate the impact of sequential sectioning of AC and CC ligaments on GH kinematics throughout an abduction motion in the coronal plane. Following an examination of the baseline, uninjured kinematics, the AC ligaments were then sectioned sequentially: (1) Anterior, (2) Inferior, (3) Posterior, and (4) Superior. Continued sectioning of CC ligamentous structures followed: the (5) trapezoid and then the (6) conoid ligaments. For each group, the GH translation and the area under the curve (AUC) were measured during abduction using an intact cadaveric shoulder. Total translation was calculated for each condition between ABD 30° and ABD 150° using the distance formula, and a univariate analysis was used to compare total translation for each axis during the different conditions. RESULTS: GH kinematics were not altered following sequential resection of the AC ligaments. Disruption of the trapezoid resulted in significant anterior and lateral displacement of the center of GH rotation. Sectioning the conoid ligament further increased the inferior shift in GH displacement. CONCLUSION: A combined injury of the AC and CC ligaments significantly alters GH kinematics during abduction. Type III AC separations, result in a significant change in the shoulder's motion and may warrant surgical reconstruction to restore normal function.

The effect of the rotator interval on glenohumeral kinematics during abduction.

BACKGROUND: The rotator interval (RI) has been exploited as a potentially benign point of entry into the glenohumeral (GH) joint. Bounded by the supraspinatus, subscapularis and coracoid process of the scapula, the RI is believed to be important in the shoulder's soft tissue balancing and function. However, the role of the RI in shoulder kinematics is not fully understood. The purpose of this study is to describe the effect of the RI on GH motion during abduction of the arm. METHODS: Six shoulders from three cadaveric torsos were studied to assess the impact of changes in the RI during abduction under four conditions: Intact (Baseline), Opened, Repaired (repaired with side-to-side tissue approximation, no overlap) and Tightened (repaired with 1 cm overlap). For each group, the GH translation and area under the Curve (AUC) were measured during abduction using an intact cadaveric shoulder (intact torso). RESULTS: GH kinematics varied in response to each intervention and throughout the entire abduction arc. Opening the RI caused a significant change in GH translation. The Repair and Tightened groups behaved similarly along all axes of GH motion. CONCLUSIONS: The RI is central to normal GH kinematics. Any insult to the tissue's integrity alters the shoulder's motion throughout abduction. In this model, closing the RI side-to-side has the same effect as tightening the RI. Since suture closure may offer the same benefit as tightening the RI, clinicians should consider this effect when treating patients with shoulder laxity. This investigation provides an improved perspective on the role of the RI on GH kinematics during abduction. When managing shoulder pathology, surgeons should consider how these different methods of RI closure affect the joint's motion. In different circumstances, the surgical approach to the RI can be tailored to address each patient's specific needs.

Biomechanics and mechanobiology of trabecular bone: a review.

Trabecular bone is a highly porous, heterogeneous, and anisotropic material which can be found at the epiphyses of long bones and in the vertebral bodies. Studying the mechanical properties of trabecular bone is important, since trabecular bone is the main load bearing bone in vertebral bodies and also transfers the load from joints to the compact bone of the cortex of long bones. This review article highlights the high dependency of the mechanical properties of trabecular bone on species, age, anatomic site, loading direction, and size of the sample under consideration. In recent years, high resolution micro finite element methods have been extensively used to specifically address the mechanical properties of the trabecular bone and provide unique tools to interpret and model the mechanical testing experiments. The aims of the current work are to first review the mechanobiology of trabecular bone and then present classical and new approaches for modeling and analyzing the trabecular bone microstructure and macrostructure and corresponding mechanical properties such as elastic properties and strength.

Effects of different loading patterns on the trabecular bone morphology of the proximal femur using adaptive bone remodeling.

In this study, the changes in the bone density of human femur model as a result of different loadings were investigated. The model initially consisted of a solid shell representing cortical bone encompassing a cubical network of interconnected rods representing trabecular bone. A computationally efficient program was developed that iteratively changed the structure of trabecular bone by keeping the local stress in the structure within a defined stress range. The stress was controlled by either enhancing existing beam elements or removing beams from the initial trabecular frame structure. Analyses were performed for two cases of homogenous isotropic and transversely isotropic beams.Trabecular bone structure was obtained for three load cases: walking, stair climbing and stumbling without falling. The results indicate that trabecular bone tissue material properties do not have a significant effect on the converged structure of trabecular bone. In addition, as the magnitude of the loads increase, the internal structure becomes denser in critical zones. Loading associated with the stumbling results in the highest density;whereas walking, considered as a routine daily activity, results in the least internal density in different regions. Furthermore, bone volume fraction at the critical regions of the converged structure is in good agreement with previously measured data obtained from combinations of dual X-ray absorptiometry (DXA) and computed tomography (CT). The results indicate that the converged bone architecture consisting of rods and plates are consistent with the natural bone morphology of the femur. The proposed model shows a promising means to understand the effects of different individual loading patterns on the bone density.

Optimal fractal-like hierarchical honeycombs.

Hexagonal honeycomb structures are known for their high strength and low weight. We construct a new class of fractal-appearing cellular metamaterials by replacing each three-edge vertex of a base hexagonal network with a smaller hexagon and iterating this process. The mechanical properties of the structure after different orders of the iteration are optimized. We find that the optimal structure (with highest in-plane stiffness for a given weight ratio) is self-similar but requires higher order hierarchy as the density vanishes. These results offer insights into how incorporating hierarchy in the material structure can create low-density metamaterials with desired properties and function.

Anisotropic blistering instability of highly ellipsoidal shells.

The formation of localized periodic structures in the deformation of elastic shells is well documented and is a familiar first stage in the crushing of a spherical shell such as a ping-pong ball. While spherical shells manifest such periodic structures as polygons, we present a new instability that is observed in the indentation of a highly ellipsoidal shell by a horizontal plate. Above a critical indentation depth, the plate loses contact with the shell in a series of well-defined "blisters" along the long axis of the ellipsoid. We characterize the onset of this instability and explain it using scaling arguments, numerical simulations, and experiments. We also characterize the properties of the blistering pattern by showing how the number of blisters and their size depend on both the geometrical properties of the shell and the indentation but not on the shell's elastic modulus. This blistering instability may be used to determine the thickness of highly ellipsoidal shells simply by squashing them between two plates.

An Automated Vertebrae Localization, Segmentation, and Osteoporotic Compression Fracture Detection Pipeline for Computed Tomographic Imaging.

Osteoporosis is the most common chronic metabolic bone disease worldwide. Vertebral compression fracture (VCF) is the most common type of osteoporotic fracture. Approximately 700,000 osteoporotic VCFs are diagnosed annually in the USA alone, resulting in an annual economic burden of ~$13.8B. With an aging population, the rate of osteoporotic VCFs and their associated burdens are expected to rise. Those burdens include pain, functional impairment, and increased medical expenditure. Therefore, it is of utmost importance to develop an analytical tool to aid in the identification of VCFs. Computed Tomography (CT) imaging is commonly used to detect occult injuries. Unlike the existing VCF detection approaches based on CT, the standard clinical criteria for determining VCF relies on the shape of vertebrae, such as loss of vertebral body height. We developed a novel automated vertebrae localization, segmentation, and osteoporotic VCF detection pipeline for CT scans using state-of-the-art deep learning models to bridge this gap. To do so, we employed a publicly available dataset of spine CT scans with 325 scans annotated for segmentation, 126 of which also graded for VCF (81 with VCFs and 45 without VCFs). Our approach attained 96% sensitivity and 81% specificity in detecting VCF at the vertebral-level, and 100% accuracy at the subject-level, outperforming deep learning counterparts tested for VCF detection without segmentation. Crucially, we showed that adding predicted vertebrae segments as inputs significantly improved VCF detection at both vertebral and subject levels by up to 14% Sensitivity and 20% Specificity (p-value = 0.028).

Detection and Localization of Spine Disorders from Plain Radiography.

Spine disorders can cause severe functional limitations, including back pain, decreased pulmonary function, and increased mortality risk. Plain radiography is the first-line imaging modality to diagnose suspected spine disorders. Nevertheless, radiographical appearance is not always sufficient due to highly variable patient and imaging parameters, which can lead to misdiagnosis or delayed diagnosis. Employing an accurate automated detection model can alleviate the workload of clinical experts, thereby reducing human errors, facilitating earlier detection, and improving diagnostic accuracy. To this end, deep learning-based computer-aided diagnosis (CAD) tools have significantly outperformed the accuracy of traditional CAD software. Motivated by these observations, we proposed a deep learning-based approach for end-to-end detection and localization of spine disorders from plain radiographs. In doing so, we took the first steps in employing state-of-the-art transformer networks to differentiate images of multiple spine disorders from healthy counterparts and localize the identified disorders, focusing on vertebral compression fractures (VCF) and spondylolisthesis due to their high prevalence and potential severity. The VCF dataset comprised 337 images, with VCFs collected from 138 subjects and 624 normal images collected from 337 subjects. The spondylolisthesis dataset comprised 413 images, with spondylolisthesis collected from 336 subjects and 782 normal images collected from 413 subjects. Transformer-based models exhibited 0.97 Area Under the Receiver Operating Characteristic Curve (AUC) in VCF detection and 0.95 AUC in spondylolisthesis detection. Further, transformers demonstrated significant performance improvements against existing end-to-end approaches by 4-14% AUC (p-values < 10-13) for VCF detection and by 14-20% AUC (p-values < 10-9) for spondylolisthesis detection.