Machine-Learning-Based Multi-Modal Force Estimation for Steerable Ablation Catheters

Catheter-based cardiac ablation is a minimally invasive procedure for treating atrial fibrillation (AF). Electrophysiologists perform the procedure under image guidance during which the contact force between the heart tissue and the catheter tip determines the quality of lesions created. This paper describes a novel multi-modal contact force estimator based on Convolutional Neural Networks (CNNs) and Recurrent Neural Networks (RNNs). The estimator takes the shape and optical flow of the deflectable distal section as two modalities since frames and motion between frames complement each other to capture the long context in the video frames of the catheter. The angle between the tissue and the catheter tip is considered a complement of the extracted shape. The data acquisition platform measures the two-degrees-of-freedom contact force and video data as the catheter motion is constrained in the imaging plane. The images are captured via a camera that simulates single-view fluoroscopy for experimental purposes. In this sensor-free procedure, the features of the images and optical flow modalities are extracted through transfer learning. Long Short-Term Memory Networks (LSTMs) with a memory fusion network (MFN) are implemented to consider time dependency and hysteresis due to friction. The architecture integrates spatial and temporal networks. Late fusion with the concatenation of LSTMs, transformer decoders, and Gated Recurrent Units (GRUs) are implemented to verify the feasibility of the proposed network-based approach and its superiority over single-modality networks. The resulting mean absolute error, which accounted for only 2.84% of the total magnitude, was obtained by collecting data under more realistic circumstances in contrast to previous research studies. The decrease in error is considerably better than that achieved by individual modalities and late fusion with concatenation. These results emphasize the practicality and relevance of utilizing a multimodal network in real-world scenarios.

Comprehensive whole-genome sequence analyses provide insights into the genomic architecture of cerebral palsy

We performed whole-genome sequencing (WGS) in 327 children with cerebral palsy (CP) and their biological parents. We classified 37 of 327 (11.3%) children as having pathogenic/likely pathogenic (P/LP) variants and 58 of 327 (17.7%) as having variants of uncertain significance. Multiple classes of P/LP variants included single-nucleotide variants (SNVs)/indels (6.7%), copy number variations (3.4%) and mitochondrial mutations (1.5%). The COL4A1 gene had the most P/LP SNVs. We also analyzed two pediatric control cohorts (n = 203 trios and n = 89 sib-pair families) to provide a baseline for de novo mutation rates and genetic burden analyses, the latter of which demonstrated associations between de novo deleterious variants and genes related to the nervous system. An enrichment analysis revealed previously undescribed plausible candidate CP genes (SMOC1, KDM5B, BCL11A and CYP51A1). A multifactorial CP risk profile and substantial presence of P/LP variants combine to support WGS in the diagnostic work-up across all CP and related phenotypes.

High-quality read-based phasing of cystic fibrosis cohort informs genetic understanding of disease modification

Phasing of heterozygous alleles is critical for interpretation of cis-effects of disease-relevant variation. We sequenced 477 individuals with cystic fibrosis (CF) using linked-read sequencing, which display an average phase block N50 of 4.39 Mb. We use these samples to construct a graph representation of CFTR haplotypes, demonstrating its utility for understanding complex CF alleles. These are visualized in a Web app, CFTbaRcodes, that enables interactive exploration of CFTR haplotypes present in this cohort. We perform fine-mapping and phasing of the chr7q35 trypsinogen locus associated with CF meconium ileus, an intestinal obstruction at birth associated with more severe CF outcomes and pancreatic disease. A 20-kb deletion polymorphism and a PRSS2 missense variant p.Thr8Ile (rs62473563) are shown to independently contribute to meconium ileus risk (p = 0.0028, p = 0.011, respectively) and are PRSS2 pancreas eQTLs (p = 9.5 × 10⁻⁷ and p = 1.4 × 10⁻⁴, respectively), suggesting the mechanism by which these polymorphisms contribute to CF. The phase information from linked reads provides a putative causal explanation for variation at a CF-relevant locus, which also has implications for the genetic basis of non-CF pancreatitis, to which this locus has been reported to contribute.

Genomic architecture of autism from comprehensive whole-genome sequence annotation

Fully understanding autism spectrum disorder (ASD) genetics requires whole-genome sequencing (WGS). We present the latest release of the Autism Speaks MSSNG resource, which includes WGS data from 5,100 individuals with ASD and 6,212 non-ASD parents and siblings (total n = 11,312). Examining a wide variety of genetic variants in MSSNG and the Simons Simplex Collection (SSC; n = 9,205), we identified ASD-associated rare variants in 718/5,100 individuals with ASD from MSSNG (14.1%) and 350/2,419 from SSC (14.5%). Considering genomic architecture, 52% were nuclear sequence-level variants, 46% were nuclear structural variants (including copy-number variants, inversions, large insertions, uniparental isodisomies, and tandem repeat expansions), and 2% were mitochondrial variants. Our study provides a guidebook for exploring genotype-phenotype correlations in families who carry ASD-associated rare variants and serves as an entry point to the expanded studies required to dissect the etiology in the ∼85% of the ASD population that remain idiopathic.

A large data resource of genomic copy number variation across neurodevelopmental disorders

Copy number variations (CNVs) are implicated across many neurodevelopmental disorders (NDDs) and contribute to their shared genetic etiology. Multiple studies have attempted to identify shared etiology among NDDs, but this is the first genome-wide CNV analysis across autism spectrum disorder (ASD), attention deficit hyperactivity disorder (ADHD), schizophrenia (SCZ), and obsessive-compulsive disorder (OCD) at once. Using microarray (Affymetrix CytoScan HD), we genotyped 2,691 subjects diagnosed with an NDD (204 SCZ, 1,838 ASD, 427 ADHD and 222 OCD) and 1,769 family members, mainly parents. We identified rare CNVs, defined as those found in <0.1% of 10,851 population control samples. We found clinically relevant CNVs (broadly defined) in 284 (10.5%) of total subjects, including 22 (10.8%) among subjects with SCZ, 209 (11.4%) with ASD, 40 (9.4%) with ADHD, and 13 (5.6%) with OCD. Among all NDD subjects, we identified 17 (0.63%) with aneuploidies and 115 (4.3%) with known genomic disorder variants. We searched further for genes impacted by different CNVs in multiple disorders. Examples of NDD-associated genes linked across more than one disorder (listed in order of occurrence frequency) are NRXN1, SEH1L, LDLRAD4, GNAL, GNG13, MKRN1, DCTN2, KNDC1, PCMTD2, KIF5A, SYNM, and long non-coding RNAs: AK127244 and PTCHD1-AS. We demonstrated that CNVs impacting the same genes could potentially contribute to the etiology of multiple NDDs. The CNVs identified will serve as a useful resource for both research and diagnostic laboratories for prioritization of variants.

The Personal Genome Project Canada: findings from whole genome sequences of the inaugural 56 participants

BACKGROUND

The Personal Genome Project Canada is a comprehensive public data resource that integrates whole genome sequencing data and health information. We describe genomic variation identified in the initial recruitment cohort of 56 volunteers.

METHODS

Volunteers were screened for eligibility and provided informed consent for open data sharing. Using blood DNA, we performed whole genome sequencing and identified all possible classes of DNA variants. A genetic counsellor explained the implication of the results to each participant.

RESULTS

Whole genome sequencing of the first 56 participants identified 207 662 805 sequence variants and 27 494 copy number variations. We analyzed a prioritized disease-associated data set (n = 1606 variants) according to standardized guidelines, and interpreted 19 variants in 14 participants (25%) as having obvious health implications. Six of these variants (e.g., in BRCA1 or mosaic loss of an X chromosome) were pathogenic or likely pathogenic. Seven were risk factors for cancer, cardiovascular or neurobehavioural conditions. Four other variants - associated with cancer, cardiac or neurodegenerative phenotypes - remained of uncertain significance because of discrepancies among databases. We also identified a large structural chromosome aberration and a likely pathogenic mitochondrial variant. There were 172 recessive disease alleles (e.g., 5 individuals carried mutations for cystic fibrosis). Pharmacogenomics analyses revealed another 3.9 potentially relevant genotypes per individual.

INTERPRETATION

Our analyses identified a spectrum of genetic variants with potential health impact in 25% of participants. When also considering recessive alleles and variants with potential pharmacologic relevance, all 56 participants had medically relevant findings. Although access is mostly limited to research, whole genome sequencing can provide specific and novel information with the potential of major impact for health care.

Whole genome sequencing resource identifies 18 new candidate genes for autism spectrum disorder

We are performing whole-genome sequencing of families with autism spectrum disorder (ASD) to build a resource (MSSNG) for subcategorizing the phenotypes and underlying genetic factors involved. Here we report sequencing of 5,205 samples from families with ASD, accompanied by clinical information, creating a database accessible on a cloud platform and through a controlled-access internet portal. We found an average of 73.8 de novo single nucleotide variants and 12.6 de novo insertions and deletions or copy number variations per ASD subject. We identified 18 new candidate ASD-risk genes and found that participants bearing mutations in susceptibility genes had significantly lower adaptive ability (P = 6 × 10^-4). In 294 of 2,620 (11.2%) of ASD cases, a molecular basis could be determined and 7.2% of these carried copy number variations and/or chromosomal abnormalities, emphasizing the importance of detecting all forms of genetic variation as diagnostic and therapeutic targets in ASD.

A Human "eFP" Browser for Generating Gene Expression Anatograms

Transcriptomic studies help to further our understanding of gene function. Human transcriptomic studies tend to focus on a particular subset of tissue types or a particular disease state; however, it is possible to collate into a compendium multiple studies that have been profiled using the same expression analysis platform to provide an overview of gene expression levels in many different tissues or under different conditions. In order to increase the knowledge and understanding we gain from such studies, intuitive visualization of gene expression data in such a compendium can be useful. The Human eFP (“electronic Fluorescent Pictograph”) Browser presented here is a tool for intuitive visualization of large human gene expression data sets on pictographic representations of the human body as gene expression “anatograms”. Pictographic representations for new data sets may be generated easily. The Human eFP Browser can also serve as a portal to other gene-specific information through link-outs to various online resources.

Inference of Longevity-Related Genes from a Robust Coexpression Network of Seed Maturation Identifies Regulators Linking Seed Storability to Biotic Defense-Related Pathways

Seed longevity, the maintenance of viability during storage, is a crucial factor for preservation of genetic resources and ensuring proper seedling establishment and high crop yield. We used a systems biology approach to identify key genes regulating the acquisition of longevity during seed maturation of Medicago truncatula. Using 104 transcriptomes from seed developmental time courses obtained in five growth environments, we generated a robust, stable coexpression network (MatNet), thereby capturing the conserved backbone of maturation. Using a trait-based gene significance measure, a coexpression module related to the acquisition of longevity was inferred from MatNet. Comparative analysis of the maturation processes in M. truncatula and Arabidopsis thaliana seeds and mining Arabidopsis interaction databases revealed conserved connectivity for 87% of longevity module nodes between both species. Arabidopsis mutant screening for longevity and maturation phenotypes demonstrated high predictive power of the longevity cross-species network. Overrepresentation analysis of the network nodes indicated biological functions related to defense, light, and auxin. Characterization of defense-related wrky3 and nf-x1-like1 (nfxl1) transcription factor mutants demonstrated that these genes regulate some of the network nodes and exhibit impaired acquisition of longevity during maturation. These data suggest that seed longevity evolved by co-opting existing genetic pathways regulating the activation of defense against pathogens.

Developmental transcriptional profiling reveals key insights into Triticeae reproductive development

Despite their importance, there remains a paucity of large-scale gene expression-based studies of reproductive development in species belonging to the Triticeae. As a first step to address this deficiency, a gene expression atlas of triticale reproductive development was generated using the 55K Affymetrix GeneChip(®) wheat genome array. The global transcriptional profiles of the anther/pollen, ovary and stigma were analyzed at concurrent developmental stages, and co-expressed as well as preferentially expressed genes were identified. Data analysis revealed both novel and conserved regulatory factors underlying Triticeae floral development and function. This comprehensive resource rests upon detailed gene annotations, and the expression profiles are readily accessible via a web browser.

Integration of force reflection with tactile sensing for minimally invasive robotics-assisted tumor localization

Tactile sensing and force reflection have been the subject of considerable research for tumor localization in soft-tissue palpation. The work presented in this paper investigates the relevance of force feedback (presented visually as well as directly) during tactile sensing (presented visually only) for tumor localization using an experimental setup close to one that could be applied for real robotics-assisted minimally invasive surgery. The setup is a teleoperated (master-slave) system facilitated with a state-of-the-art minimally invasive probe with a rigidly mounted tactile sensor at the tip and an externally mounted force sensor at the base of the probe. The objective is to capture the tactile information and measure the interaction forces between the probe and tissue during palpation and to explore how they can be integrated to improve the performance of tumor localization. To quantitatively explore the effect of force feedback on tactile sensing tumor localization, several experiments were conducted by human subjects to locate artificial tumors embedded in the ex vivo bovine livers. The results show that using tactile sensing in a force-controlled environment can realize, on average, 57 percent decrease in the maximum force and 55 percent decrease in the average force applied to tissue while increasing the tumor detection accuracy by up to 50 percent compared to the case of using tactile feedback alone. The results also show that while visual presentation of force feedback gives straightforward quantitative measures, improved performance of tactile sensing tumor localization is achieved at the expense of longer times for the user. Also, the quickness and intuitive data mapping of direct force feedback makes it more appealing to experienced users.

A semi-infinite programming approach to preoperative planning of robotic cardiac surgery under geometric uncertainty

In this paper, a computational framework for patient-specific preoperative planning of Robotics-Assisted Minimally Invasive Cardiac Surgery (RAMICS) is presented. It is expected that preoperative planning of RAMICS will improve the success rate by considering robot kinematics, patient-specific thoracic anatomy, and procedure-specific intraoperative conditions. Given the significant anatomical features localized in the preoperative computed tomography images of a patients thorax, port locations and robot orientations (with respect to the patients body coordinate frame) are determined to optimize qualities such as dexterity, reachability, tool approach angles and maneuverability. To address intraoperative geometric uncertainty, the problem is formulated as a Generalized Semi-Infinite Program (GSIP) with a convex lower-level problem to seek a plan that is less sensitive to geometric uncertainty in the neighborhood of surgical targets. It is demonstrated that with a proper formulation of the problem, the GSIP can be replaced by a tractable constrained nonlinear program that uses a multi-criteria objective function to balance between the nominal task performance and robustness to collisions and joint limit violations. Finally, performance of the proposed formulation is demonstrated by a comparison between the plans generated by the algorithm and those recommended by an experienced surgeon for several case studies.

BAR expressolog identification: expression profile similarity ranking of homologous genes in plant species

Large numbers of sequences are now readily available for many plant species, allowing easy identification of homologous genes. However, orthologous gene identification across multiple species is made difficult by evolutionary events such as whole-genome or segmental duplications. Several developmental atlases of gene expression have been produced in the past couple of years, and it may be possible to use these transcript abundance data to refine ortholog predictions. In this study, clusters of homologous genes between seven plant species - Arabidopsis, soybean, Medicago truncatula, poplar, barley, maize and rice - were identified. Following this, a pipeline to rank homologs within gene clusters by both sequence and expression profile similarity was devised by determining equivalent tissues between species, with the best expression profile match being termed the 'expressolog'. Five electronic fluorescent pictograph (eFP) browsers were produced as part of this effort, to aid in visualization of gene expression data and to complement existing eFP browsers at the Bio-Array Resource (BAR). Within the eFP browser framework, these expression profile similarity rankings were incorporated into an Expressolog Tree Viewer to allow cross-species homolog browsing by both sequence and expression pattern similarity. Global analyses showed that orthologs with the highest sequence similarity do not necessarily exhibit the highest expression pattern similarity. Other orthologs may show different expression patterns, indicating that such genes may require re-annotation or more specific annotation. Ultimately, it is envisaged that this pipeline will aid in improvement of the functional annotation of genes and translational plant research.

A fast torsionally compliant kinematic model of concentric-tube robots

Concentric-tube robots have the potential to become an important surgical tool for robot-assisted percutaneous interventions. They can provide dexterous operation in a small constrained environment. The kinematic model of a concentric-tube robot has been well developed in terms of accuracy, but the computational cost places limitations on real-time implementation. In this paper, we propose a new technique that will substantially improve the computational efficiency of evaluating the kinematics of a concentric-tube robot in the context of developing a control strategy without sacrificing the accuracy of the results. In this paper we develop a torsionally compliant kinematic model using global variables. The model is validated by comparing the results obtained by computing the kinematic model corresponding to an experimental setup of a concentric-tube robot to which a force/torque sensor has been mounted at its base with those obtained directly from the experimental setup. The results indicate that it is feasible to compute the kinematics of the concentric-tube robot fast enough to allow the position/force control loop to be implemented at a rate of 1 kHz.

Optimal transseptal puncture location for robot-assisted left atrial catheter ablation

BACKGROUND

The preferred method of treatment for atrial fibrillation (AF) is by catheter ablation, in which a catheter is guided into the left atrium through a transseptal puncture. However, the transseptal puncture constrains the catheter, thereby limiting its manoeuvrability and increasing the difficulty in reaching various locations in the left atrium. In this paper, we address the problem of choosing the optimal transseptal puncture location for performing cardiac ablation to obtain maximum manoeuvrability of the catheter.

METHODS

We have employed an optimization algorithm to maximize the global isotropy index (GII) to evaluate the optimal transseptal puncture location. As part of this algorithm, a novel kinematic model for the catheter has been developed, based on a continuum robot model. Pre-operative MR/CT images of the heart are segmented using the open source image-guided therapy software, 3D Slicer, to obtain models of the left atrium and septal wall. These models are input to the optimization algorithm to evaluate the optimal transseptal puncture location.

RESULTS

The continuum robot model accurately describes the kinematics of the catheter. Simulation and experimental results for the optimal transseptal puncture location are presented in this paper. The optimization algorithm generates discrete points on the septal wall for which the dexterity of the catheter in the left atrium is maximum, corresponding to a GII of 0.4362.

CONCLUSION

We have developed an optimization algorithm based on the GII to evaluate the optimal position of the transseptal puncture for left atrial cardiac ablation.

New tactile sensing system for minimally invasive surgical tumour localization

BACKGROUND

Minimally invasive surgery (MIS) suffers from the inability to directly palpate organs for tumour localization. A tactile sensing system (TSS), consisting of a probe and a visualization interface, was developed to present an active pressure map of the contact surface to locate tumours during MIS.

METHODS

The TSS performance was compared to MIS graspers to locate occult 10 mm phantom tumours in ex vivo bovine liver and ex vivo porcine lung. Performance assessment included applied pressure, localization distance and accuracy.

RESULTS

The TSS realized a relative 71% reduction in maximum applied pressure and a 31% increase in detection accuracy in liver tissue (when compared to MIS graspers) and demonstrated no significant differences in performance when palpating lung tissue.

CONCLUSIONS

The TSS may help surgeons to identify occult tumours during surgery by restoring some of the haptic information lost during MIS.

Electromagnetic navigation improves minimally invasive robot-assisted lung brachytherapy

OBJECTIVE

Recent advances in minimally invasive thoracic surgery have renewed an interest in the role of interstitial brachytherapy for lung cancer. Our previous work has demonstrated that a minimally invasive robot-assisted (MIRA) lung brachytherapy system produced results that were equal to or better than those obtained with standard video-assisted thoracic surgery (VATS) and comparable to results with open surgery. The purpose of this project was to evaluate the performance of an integrated system for MIRA lung brachytherapy that incorporated modified electromagnetic navigation and ultrasound image guidance with robotic assistance.

METHODS

The experimental test-bed consisted of a VATS box, ZEUS and AESOP surgical robotic arms, a seed injector, an ultrasound machine, video monitors, a computer, and an endoscope. Our previous custom-designed electromagnetic navigational software and the robotic controller were modified and incorporated into the MIRA III system to become the next-generation MIRA IV. Inactive brachytherapy seeds were injected as close as possible to a small metal ball target embedded in an opaque agar cube. The completion time, the number of attempts, and the accuracy of seed deployment were compared for manual placement, standard VATS, MIRA III, and the new MIRA IV system.

RESULTS

The MIRA IV system significantly reduced the median procedure time by 61% (104 s to 41 s), tissue trauma by 75% (4 attempts to 1 attempt), and mean seed placement error by 64% (2.5 mm to 0.9 mm) when compared to a standard VATS. MIRA IV also reduced the mean procedure time by 48% (85 s to 44 s) and the seed placement error by 68% (2.8 mm to 0.9 mm) compared to the MIRA III system.

CONCLUSIONS

A modified integrated system for performing minimally invasive robot-assisted lung brachytherapy was developed that incorporated electromagnetic navigation and an improved robotic controller. The MIRA IV system performed significantly better than standard VATS and better than MIRA III.

Tool/tissue interaction feedback modalities in robot-assisted lump localization

Providing a surgeon with information regarding contacts made between tools and tissue during robot-assisted interventions can improve task efficiency and reliability. It is hypothesized that various modalities of contact feedback have the potential to enhance performance in a robot-assisted minimally invasive environment. In this paper, (kinesthetic) haptic feedback is compared with visual feedback of haptic information in terms of several performance metrics. Using a haptics-capable master-slave test-bed for endoscopic surgery, experiments involving a lump localization task are conducted and the performance of human subjects is compared for these two modalities of contact feedback. It is shown that the two feedback modalities result in comparable localization accuracies--an advantage of visual haptic feedback due to the lower system complexity required--while the task completion times are significantly shorter with haptic feedback.