1
|
Fan Y, Zhang D, Banalagay R, Wang J, Noble JH, Dawant BM. Hybrid active shape and deep learning method for the accurate and robust segmentation of the intracochlear anatomy in clinical head CT and CBCT images. J Med Imaging (Bellingham) 2021; 8:064002. [PMID: 34853805 DOI: 10.1117/1.jmi.8.6.064002] [Citation(s) in RCA: 1] [Impact Index Per Article: 0.3] [Reference Citation Analysis] [Abstract] [Key Words] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Received: 12/27/2020] [Accepted: 11/08/2021] [Indexed: 11/14/2022] Open
Abstract
Purpose: Robust and accurate segmentation methods for the intracochlear anatomy (ICA) are a critical step in the image-guided cochlear implant programming process. We have proposed an active shape model (ASM)-based method and a deep learning (DL)-based method for this task, and we have observed that the DL method tends to be more accurate than the ASM method while the ASM method tends to be more robust. Approach: We propose a DL-based U-Net-like architecture that incorporates ASM segmentation into the network. A quantitative analysis is performed on a dataset that consists of 11 cochlea specimens for which a segmentation ground truth is available. To qualitatively evaluate the robustness of the method, an experienced expert is asked to visually inspect and grade the segmentation results on a clinical dataset made of 138 image volumes acquired with conventional CT scanners and of 39 image volumes acquired with cone beam CT (CBCT) scanners. Finally, we compare training the network (1) first with the ASM results, and then fine-tuning it with the ground truth segmentation and (2) directly with the specimens with ground truth segmentation. Results: Quantitative and qualitative results show that the proposed method increases substantially the robustness of the DL method while having only a minor detrimental effect (though not significant) on its accuracy. Expert evaluation of the clinical dataset shows that by incorporating the ASM segmentation into the DL network, the proportion of good segmentation cases increases from 60/177 to 119/177 when training only with the specimens and increases from 129/177 to 151/177 when pretraining with the ASM results. Conclusions: A hybrid ASM and DL-based segmentation method is proposed to segment the ICA in CT and CBCT images. Our results show that combining DL and ASM methods leads to a solution that is both robust and accurate.
Collapse
Affiliation(s)
- Yubo Fan
- Vanderbilt University, Department of Computer Science, Nashville, Tennessee, United States
| | | | - Rueben Banalagay
- Vanderbilt University, Department of Electrical and Computer Engineering, Nashville, Tennessee, United States
| | - Jianing Wang
- Vanderbilt University, Department of Electrical and Computer Engineering, Nashville, Tennessee, United States
| | - Jack H Noble
- Vanderbilt University, Department of Electrical and Computer Engineering, Nashville, Tennessee, United States
| | - Benoit M Dawant
- Vanderbilt University, Department of Electrical and Computer Engineering, Nashville, Tennessee, United States
| |
Collapse
|
2
|
Torres R, Hochet B, Daoudi H, Carré F, Mosnier I, Sterkers O, Ferrary E, Nguyen Y. Atraumatic Insertion of a Cochlear Implant Pre-Curved Electrode Array by a Robot-Automated Alignment with the Coiling Direction of the Scala Tympani. Audiol Neurootol 2021; 27:148-155. [PMID: 34284383 DOI: 10.1159/000517398] [Citation(s) in RCA: 7] [Impact Index Per Article: 2.3] [Reference Citation Analysis] [Abstract] [Key Words] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Received: 12/22/2020] [Accepted: 05/24/2021] [Indexed: 11/19/2022] Open
Abstract
INTRODUCTION Electrode array translocation is an unpredictable event with all types of arrays, even using a teleoperated robot in a clinical scenario. We aimed to compare the intracochlear trauma produced by the HiFocus™ Mid-Scala (MS) electrode array (Advanced Bionics, Valencia, CA, USA) using a teleoperated robot, with an automated robot connected to a navigation system to align the pre-curved tip of the electrode array with the coiling direction of the scala tympani (ST). METHODS Fifteen freshly frozen temporal bones were implanted with the MS array using the RobOtol® (Collin, Bagneux, France). In the first group (n = 10), the robot was teleoperated to insert the electrode array into the basal turn of the ST under stereomicroscopic vision, and then the array was driven by a slow-speed hydraulic insertion technique with an estimated placement of the pre-curved electrode tip. In the second group (n = 5), 3 points were obtained from the preoperative cone-beam computed tomography: the 2 first defining the ST insertion axis of the basal turn and a third one at the center of the ST at 270°. They provided the information to the automated system (RobOtol® connected with a navigation system) to automatically align the electrode array with the ST insertion axis and to aim the pre-curved tip toward the subsequent coiling of the ST. After this, the electrode array was manually advanced. Finally, the cochleae were obtained and fixed in a crystal resin, and the position of each electrode was determined by a micro-grinding technique. RESULTS In all cases, the electrode array was fully inserted into the cochlea and the depth of insertion was similar using both techniques. With the teleoperated robotic technique, translocations of the array were observed in 7/10 insertions (70%), but neither trauma nor array translocation occurred with automated robotic insertion. CONCLUSION We have successfully tested an automated insertion system (robot + navigation) that could accurately align a pre-curved electrode array to the axis of the basal turn of the ST and its subsequent coiling, which reduced intracochlear insertion trauma and translocation.
Collapse
Affiliation(s)
- Renato Torres
- Technologies et thérapie génique pour la surdité, Institut de l'Audition, Institut Pasteur/Inserm, Paris, France.,Departamento de Ciencias Fisiológicas, Facultad de Medicina, Universidad Nacional de San Agustín de Arequipa, Arequipa, Peru
| | - Baptiste Hochet
- Technologies et thérapie génique pour la surdité, Institut de l'Audition, Institut Pasteur/Inserm, Paris, France.,Unité fonctionnelle Implants auditifs et explorations fonctionnelles, Service ORL, GH Pitié-Salpêtrière, AP-HP Sorbonne Université, Paris, France
| | - Hannah Daoudi
- Technologies et thérapie génique pour la surdité, Institut de l'Audition, Institut Pasteur/Inserm, Paris, France.,Unité fonctionnelle Implants auditifs et explorations fonctionnelles, Service ORL, GH Pitié-Salpêtrière, AP-HP Sorbonne Université, Paris, France
| | - Fabienne Carré
- Technologies et thérapie génique pour la surdité, Institut de l'Audition, Institut Pasteur/Inserm, Paris, France.,Unité fonctionnelle Implants auditifs et explorations fonctionnelles, Service ORL, GH Pitié-Salpêtrière, AP-HP Sorbonne Université, Paris, France
| | - Isabelle Mosnier
- Technologies et thérapie génique pour la surdité, Institut de l'Audition, Institut Pasteur/Inserm, Paris, France.,Unité fonctionnelle Implants auditifs et explorations fonctionnelles, Service ORL, GH Pitié-Salpêtrière, AP-HP Sorbonne Université, Paris, France
| | - Olivier Sterkers
- Technologies et thérapie génique pour la surdité, Institut de l'Audition, Institut Pasteur/Inserm, Paris, France.,Unité fonctionnelle Implants auditifs et explorations fonctionnelles, Service ORL, GH Pitié-Salpêtrière, AP-HP Sorbonne Université, Paris, France
| | - Evelyne Ferrary
- Technologies et thérapie génique pour la surdité, Institut de l'Audition, Institut Pasteur/Inserm, Paris, France.,Unité fonctionnelle Implants auditifs et explorations fonctionnelles, Service ORL, GH Pitié-Salpêtrière, AP-HP Sorbonne Université, Paris, France
| | - Yann Nguyen
- Technologies et thérapie génique pour la surdité, Institut de l'Audition, Institut Pasteur/Inserm, Paris, France.,Unité fonctionnelle Implants auditifs et explorations fonctionnelles, Service ORL, GH Pitié-Salpêtrière, AP-HP Sorbonne Université, Paris, France
| |
Collapse
|
3
|
The Insertion Results of a Mid-scala Electrode Assessed by MRI and CBCT Image Fusion. Otol Neurotol 2019; 39:e1019-e1025. [PMID: 30444847 DOI: 10.1097/mao.0000000000002045] [Citation(s) in RCA: 11] [Impact Index Per Article: 2.2] [Reference Citation Analysis] [Abstract] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Indexed: 11/25/2022]
Abstract
OBJECTIVES To investigate the results of clinical surgical insertions with a Mid-scala array (HIFocus Mid-Scala Electrode, HFms). STUDY DESIGN Consecutive retrospective case study. SETTINGS Tertiary referral center. PATIENTS Analyses of imaging data of 26 consecutive patients (31 insertions) implanted with the HFms. INTERVENTION (S) The evaluation of insertion trauma evoked by a previously validated image fusion technique. Electrode reconstructions from postoperative cone-beam computed tomography (CBCT) were overlaid onto preoperative magnetic resonance imaging (MRI) scans to create artifact-free images. MAIN OUTCOME MEASURES The electrode position was quantified in relation to the basilar membrane. Trauma scaling adopted from Eshraghi was used for evaluating insertion trauma. The results of the visual assessment of the postoperative CBCT were compared to those obtained with the fusion technique. RESULTS Three insertions had to be excluded due to incompatibility of the imaging data with the fusion software. We found consistent peri- to mid-modiolar placement of the HFms with a mean insertion depth angle of 376°. According to the medical records, a visual examination of the postoperative CBCT indicated that there had been no scala dislocations but when assessed by the image fusion technique, five scala dislocations (17.8%) were found. Additionally, one tip fold-over was detected in the postoperative CBCT even though this was not evident in any intraoperative measurements. CONCLUSION HFms showed atraumatic surgical insertion results with consistent mid-modiolar placement. Image fusion enhances the accuracy of the insertion trauma assessment. Routine postoperative imaging is recommended for identifying tip fold-over as well as for quality control and documentation.
Collapse
|
4
|
Cochlear Implantation With a Novel Long Straight Electrode: the Insertion Results Evaluated by Imaging and Histology in Human Temporal Bones. Otol Neurotol 2019; 39:e784-e793. [PMID: 30199496 DOI: 10.1097/mao.0000000000001953] [Citation(s) in RCA: 10] [Impact Index Per Article: 2.0] [Reference Citation Analysis] [Abstract] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Indexed: 11/25/2022]
Abstract
HYPOTHESIS To evaluate the insertion results of a novel straight array (EVO) by detailed imaging and subsequent histology in human temporal bones (TB). BACKGROUND The main focuses of modern cochlear implant surgery are to prevent damage to the intracochlear structures and to preserve residual hearing. This is often achievable with new atraumatic electrode arrays in combination with meticulous surgical techniques. METHODS Twenty fresh-frozen TBs were implanted with the EVO. Pre- and postoperative cone beam computed tomography scans were reconstructed and fused for an artifact-free representation of the electrode. The array's vertical position was quantified in relation to the basilar membrane on basis of which trauma was classified (Grades 0-4). The basilar membrane location was modeled from previous histologic data. The TBs underwent subsequent histologic examination. RESULTS The EVOs were successfully inserted in all TBs. Atraumatic insertion (Grades 0-1) were accomplished in 14 of 20 TBs (70%). There were three apical translocations, and two basal translocations due to electrode bulging. One TB had multiple translocations. The sensitivity and specificity of imaging for detecting insertion trauma (Grades 2-4) was 87.5% and 97.3.0%, respectively. CONCLUSION Comparable insertion results as reported for other arrays were also found for the EVO. Insertion trauma can be mostly avoided with meticulous insertion techniques to prevent bulging and by limiting the insertion depth angle to 360 degrees. The image fusion technique is a reliable tool for evaluating electrode placement and is feasible for trauma grading.
Collapse
|
5
|
Dhanasingh A. Why Pre-Curved Modiolar Hugging Electrodes Only Cover The Basal Turn of The Cochlea and Not Beyond that? J Int Adv Otol 2019; 14:376-381. [PMID: 30644378 DOI: 10.5152/iao.2018.5831] [Citation(s) in RCA: 1] [Impact Index Per Article: 0.2] [Reference Citation Analysis] [Abstract] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Indexed: 11/22/2022] Open
Abstract
The question of why pre-curved modiolar hugging (MH) electrodes only cover the basal turn of the cochlea and not beyond that is unanswered yet in the CI field. Therefore the aim of this article is to show what the practical limitations are with the pre-curved MH electrode design in not being able to fabricate beyond one full turn. Every CI electrode design needs a metal mold with grooves for placing the platinum wires and for injecting with the silicone elastomer. Limitations in making a mold with groove that goes beyond one full turn of curvature along with the mechanical deformation of the curved silicone elastomer, prevents making a pre-curved MH electrode beyond one full turn. Electrode tip fold-over, electrode scalar deviation and the inconsistent electrode to modiolus wall proximity are the reported issues with this electrode type which does not help by any means to the operating surgeon and the pediatric candidates especially. If intra-operative imaging is recommended to confirm the proper placement of the electrode for one particular electrode design, then how many clinics in the world may have this facility and is it ethical to put the patient under more radiation risk are the natural questions that needs to be answered in the interest of the patient. Every CI brand should come out of their marketing philosophy and innovate what is essential in bringing the full benefit of the device to the patients.
Collapse
|
6
|
Zhu R, Li J, Wang Z, Wu Y, Wang B, Zhao K. Multi-rigid-body modeling and simulation of perimodiolar cochlear electrode arrays. Proc Inst Mech Eng H 2019; 233:483-496. [PMID: 30895876 DOI: 10.1177/0954411919834485] [Citation(s) in RCA: 3] [Impact Index Per Article: 0.6] [Reference Citation Analysis] [Abstract] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Indexed: 11/17/2022]
Abstract
This study presented a method that decomposes perimodiolar electrodes into multi-rigid bodies for the study on the shape variation of cochlear perimodiolar electrode. The coordinates of electrode array were obtained by capturing the shape varying image of the perimodiolar electrodes with the stylet extracted. Subsequently, the increment of the angle variation and the length of each link were obtained. Fourier compensation fitting method was developed using the three fitting methods to compare and analyze the increment of the angle variation of the perimodiolar electrode multi-rigid model. This can not only ensure that the initial angle of the joint is consistent with the actual angle of the perimodiolar electrode, but also fully reflect the varying trend of the joint angle of the multi-rigid model of the perimodiolar electrode. The simulation of the shape variation of the perimodiolar electrode multi-rigid-body model was performed using this method in the ADAMS simulation platform. According to the simulation results, the precise and continuous shape variation of perimodiolar electrodes can be obtained using this method.
Collapse
Affiliation(s)
| | - Jianjun Li
- China Jiliang University, Hangzhou, China
| | - Zuo Wang
- China Jiliang University, Hangzhou, China
| | - Yongzhen Wu
- Eye & ENT Hospital of Fudan University, Shanghai, China
| | | | - Kai Zhao
- China Jiliang University, Hangzhou, China
| |
Collapse
|
7
|
A New Slim Modiolar Electrode Array for Cochlear Implantation: A Radiological and Histological Study. Otol Neurotol 2018; 38:e327-e334. [PMID: 28796083 DOI: 10.1097/mao.0000000000001542] [Citation(s) in RCA: 20] [Impact Index Per Article: 3.3] [Reference Citation Analysis] [Abstract] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Indexed: 10/19/2022]
Abstract
HYPOTHESIS To explore the results of a new slim modiolar electrode array (SMA) with respect to intracochlear placement and trauma evaluated by detailed radiologic imaging and histology. BACKGROUND Hearing and structure preservation is the goal of cochlear implantation for advanced hearing outcomes. Currently, this is most consistently achieved with thin lateral wall electrodes. Modiolar electrodes are located nearer the modiolus and may provide some electrophysiological advantages, but have a greater tendency for causing insertion trauma. METHODS The SMA was implanted in 20 fresh-frozen human temporal bones (TB). All TBs were scanned pre- and postoperatively with cone beam computed tomography. For atraumatic insertion, the round window approach was preferred. Scalar localization and trauma were analyzed by three-dimensional image fusion reconstructions of the pre- and postimplant scans. The TBs underwent histologic examination to validate the radiologic findings. RESULTS Insertion through the round window was performed in 19 TBs and through a cochleostomy in one TB. In one TB trauma in the form of scala translocation was identified radiologically and histologically. In the remaining TBs there was no insertion trauma. Adequate modiolar localization of the SMA was found in 19 of 20 TBs. The mean angular insertion depth was 400 degrees without correlation to cochlea size. There was no significant statistical difference between the radiological and histological measurements of electrode localization. CONCLUSION The SMA showed consistent and atraumatic insertion results in TBs. Pre- and postimplant cone beam computed tomography with image fusion was shown to be very accurate for the assessment of electrode position and insertion trauma.
Collapse
|