Intraoperative 2D/3D Image Registration via Differentiable X-ray Rendering

Surgical decisions are informed by aligning rapid portable 2D intraoperative images (e.g., X-rays) to a high-fidelity 3D preoperative reference scan (e.g., CT). 2D/3D image registration often fails in practice: conventional optimization methods are prohibitively slow and susceptible to local minima,...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Gopalakrishnan, Vivek, Dey, Neel, Golland, Polina
Format: Artikel
Sprache:eng
Schlagworte:
Online-Zugang:Volltext bestellen
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
container_end_page
container_issue
container_start_page
container_title
container_volume
creator Gopalakrishnan, Vivek
Dey, Neel
Golland, Polina
description Surgical decisions are informed by aligning rapid portable 2D intraoperative images (e.g., X-rays) to a high-fidelity 3D preoperative reference scan (e.g., CT). 2D/3D image registration often fails in practice: conventional optimization methods are prohibitively slow and susceptible to local minima, while neural networks trained on small datasets fail on new patients or require impractical landmark supervision. We present DiffPose, a self-supervised approach that leverages patient-specific simulation and differentiable physics-based rendering to achieve accurate 2D/3D registration without relying on manually labeled data. Preoperatively, a CNN is trained to regress the pose of a randomly oriented synthetic X-ray rendered from the preoperative CT. The CNN then initializes rapid intraoperative test-time optimization that uses the differentiable X-ray renderer to refine the solution. Our work further proposes several geometrically principled methods for sampling camera poses from $\mathbf{SE}(3)$, for sparse differentiable rendering, and for driving registration in the tangent space $\mathfrak{se}(3)$ with geodesic and multiscale locality-sensitive losses. DiffPose achieves sub-millimeter accuracy across surgical datasets at intraoperative speeds, improving upon existing unsupervised methods by an order of magnitude and even outperforming supervised baselines. Our code is available at https://github.com/eigenvivek/DiffPose.
doi_str_mv 10.48550/arxiv.2312.06358
format Article
fullrecord <record><control><sourceid>arxiv_GOX</sourceid><recordid>TN_cdi_arxiv_primary_2312_06358</recordid><sourceformat>XML</sourceformat><sourcesystem>PC</sourcesystem><sourcerecordid>2312_06358</sourcerecordid><originalsourceid>FETCH-LOGICAL-a678-8ce0c16bd4c3ba6e4b6b7b17fa5214de071756eba06ed0bee5f5d74f7d056f23</originalsourceid><addsrcrecordid>eNotz7tuwjAYBWAvHSroA3TCL5Dgu7Mi0ksqJCRgYIt-x78jS2CQiaLy9uXS6Qzn6EgfIe-clarSms0h_8axFJKLkhmpq1fy06Qhw-mMGYY4IhX1XNa0OUKPdIN9vAz34pToGIHWMQTMmIYI7oB0X2S43lbJY46pn5KXAIcLvv3nhGw_P3bL72K1_mqWi1UBxlZF1SHruHFeddKBQeWMs47bAFpw5ZFZbrVBB8ygZw5RB-2tCtYzbYKQEzJ7vj4s7TnHI-Rreze1D5P8A9BuR2s</addsrcrecordid><sourcetype>Open Access Repository</sourcetype><iscdi>true</iscdi><recordtype>article</recordtype></control><display><type>article</type><title>Intraoperative 2D/3D Image Registration via Differentiable X-ray Rendering</title><source>arXiv.org</source><creator>Gopalakrishnan, Vivek ; Dey, Neel ; Golland, Polina</creator><creatorcontrib>Gopalakrishnan, Vivek ; Dey, Neel ; Golland, Polina</creatorcontrib><description>Surgical decisions are informed by aligning rapid portable 2D intraoperative images (e.g., X-rays) to a high-fidelity 3D preoperative reference scan (e.g., CT). 2D/3D image registration often fails in practice: conventional optimization methods are prohibitively slow and susceptible to local minima, while neural networks trained on small datasets fail on new patients or require impractical landmark supervision. We present DiffPose, a self-supervised approach that leverages patient-specific simulation and differentiable physics-based rendering to achieve accurate 2D/3D registration without relying on manually labeled data. Preoperatively, a CNN is trained to regress the pose of a randomly oriented synthetic X-ray rendered from the preoperative CT. The CNN then initializes rapid intraoperative test-time optimization that uses the differentiable X-ray renderer to refine the solution. Our work further proposes several geometrically principled methods for sampling camera poses from $\mathbf{SE}(3)$, for sparse differentiable rendering, and for driving registration in the tangent space $\mathfrak{se}(3)$ with geodesic and multiscale locality-sensitive losses. DiffPose achieves sub-millimeter accuracy across surgical datasets at intraoperative speeds, improving upon existing unsupervised methods by an order of magnitude and even outperforming supervised baselines. Our code is available at https://github.com/eigenvivek/DiffPose.</description><identifier>DOI: 10.48550/arxiv.2312.06358</identifier><language>eng</language><subject>Computer Science - Computer Vision and Pattern Recognition</subject><creationdate>2023-12</creationdate><rights>http://creativecommons.org/licenses/by/4.0</rights><oa>free_for_read</oa><woscitedreferencessubscribed>false</woscitedreferencessubscribed></display><links><openurl>$$Topenurl_article</openurl><openurlfulltext>$$Topenurlfull_article</openurlfulltext><thumbnail>$$Tsyndetics_thumb_exl</thumbnail><link.rule.ids>228,230,780,885</link.rule.ids><linktorsrc>$$Uhttps://arxiv.org/abs/2312.06358$$EView_record_in_Cornell_University$$FView_record_in_$$GCornell_University$$Hfree_for_read</linktorsrc><backlink>$$Uhttps://doi.org/10.48550/arXiv.2312.06358$$DView paper in arXiv$$Hfree_for_read</backlink></links><search><creatorcontrib>Gopalakrishnan, Vivek</creatorcontrib><creatorcontrib>Dey, Neel</creatorcontrib><creatorcontrib>Golland, Polina</creatorcontrib><title>Intraoperative 2D/3D Image Registration via Differentiable X-ray Rendering</title><description>Surgical decisions are informed by aligning rapid portable 2D intraoperative images (e.g., X-rays) to a high-fidelity 3D preoperative reference scan (e.g., CT). 2D/3D image registration often fails in practice: conventional optimization methods are prohibitively slow and susceptible to local minima, while neural networks trained on small datasets fail on new patients or require impractical landmark supervision. We present DiffPose, a self-supervised approach that leverages patient-specific simulation and differentiable physics-based rendering to achieve accurate 2D/3D registration without relying on manually labeled data. Preoperatively, a CNN is trained to regress the pose of a randomly oriented synthetic X-ray rendered from the preoperative CT. The CNN then initializes rapid intraoperative test-time optimization that uses the differentiable X-ray renderer to refine the solution. Our work further proposes several geometrically principled methods for sampling camera poses from $\mathbf{SE}(3)$, for sparse differentiable rendering, and for driving registration in the tangent space $\mathfrak{se}(3)$ with geodesic and multiscale locality-sensitive losses. DiffPose achieves sub-millimeter accuracy across surgical datasets at intraoperative speeds, improving upon existing unsupervised methods by an order of magnitude and even outperforming supervised baselines. Our code is available at https://github.com/eigenvivek/DiffPose.</description><subject>Computer Science - Computer Vision and Pattern Recognition</subject><fulltext>true</fulltext><rsrctype>article</rsrctype><creationdate>2023</creationdate><recordtype>article</recordtype><sourceid>GOX</sourceid><recordid>eNotz7tuwjAYBWAvHSroA3TCL5Dgu7Mi0ksqJCRgYIt-x78jS2CQiaLy9uXS6Qzn6EgfIe-clarSms0h_8axFJKLkhmpq1fy06Qhw-mMGYY4IhX1XNa0OUKPdIN9vAz34pToGIHWMQTMmIYI7oB0X2S43lbJY46pn5KXAIcLvv3nhGw_P3bL72K1_mqWi1UBxlZF1SHruHFeddKBQeWMs47bAFpw5ZFZbrVBB8ygZw5RB-2tCtYzbYKQEzJ7vj4s7TnHI-Rreze1D5P8A9BuR2s</recordid><startdate>20231211</startdate><enddate>20231211</enddate><creator>Gopalakrishnan, Vivek</creator><creator>Dey, Neel</creator><creator>Golland, Polina</creator><scope>AKY</scope><scope>GOX</scope></search><sort><creationdate>20231211</creationdate><title>Intraoperative 2D/3D Image Registration via Differentiable X-ray Rendering</title><author>Gopalakrishnan, Vivek ; Dey, Neel ; Golland, Polina</author></sort><facets><frbrtype>5</frbrtype><frbrgroupid>cdi_FETCH-LOGICAL-a678-8ce0c16bd4c3ba6e4b6b7b17fa5214de071756eba06ed0bee5f5d74f7d056f23</frbrgroupid><rsrctype>articles</rsrctype><prefilter>articles</prefilter><language>eng</language><creationdate>2023</creationdate><topic>Computer Science - Computer Vision and Pattern Recognition</topic><toplevel>online_resources</toplevel><creatorcontrib>Gopalakrishnan, Vivek</creatorcontrib><creatorcontrib>Dey, Neel</creatorcontrib><creatorcontrib>Golland, Polina</creatorcontrib><collection>arXiv Computer Science</collection><collection>arXiv.org</collection></facets><delivery><delcategory>Remote Search Resource</delcategory><fulltext>fulltext_linktorsrc</fulltext></delivery><addata><au>Gopalakrishnan, Vivek</au><au>Dey, Neel</au><au>Golland, Polina</au><format>journal</format><genre>article</genre><ristype>JOUR</ristype><atitle>Intraoperative 2D/3D Image Registration via Differentiable X-ray Rendering</atitle><date>2023-12-11</date><risdate>2023</risdate><abstract>Surgical decisions are informed by aligning rapid portable 2D intraoperative images (e.g., X-rays) to a high-fidelity 3D preoperative reference scan (e.g., CT). 2D/3D image registration often fails in practice: conventional optimization methods are prohibitively slow and susceptible to local minima, while neural networks trained on small datasets fail on new patients or require impractical landmark supervision. We present DiffPose, a self-supervised approach that leverages patient-specific simulation and differentiable physics-based rendering to achieve accurate 2D/3D registration without relying on manually labeled data. Preoperatively, a CNN is trained to regress the pose of a randomly oriented synthetic X-ray rendered from the preoperative CT. The CNN then initializes rapid intraoperative test-time optimization that uses the differentiable X-ray renderer to refine the solution. Our work further proposes several geometrically principled methods for sampling camera poses from $\mathbf{SE}(3)$, for sparse differentiable rendering, and for driving registration in the tangent space $\mathfrak{se}(3)$ with geodesic and multiscale locality-sensitive losses. DiffPose achieves sub-millimeter accuracy across surgical datasets at intraoperative speeds, improving upon existing unsupervised methods by an order of magnitude and even outperforming supervised baselines. Our code is available at https://github.com/eigenvivek/DiffPose.</abstract><doi>10.48550/arxiv.2312.06358</doi><oa>free_for_read</oa></addata></record>
fulltext fulltext_linktorsrc
identifier DOI: 10.48550/arxiv.2312.06358
ispartof
issn
language eng
recordid cdi_arxiv_primary_2312_06358
source arXiv.org
subjects Computer Science - Computer Vision and Pattern Recognition
title Intraoperative 2D/3D Image Registration via Differentiable X-ray Rendering
url https://sfx.bib-bvb.de/sfx_tum?ctx_ver=Z39.88-2004&ctx_enc=info:ofi/enc:UTF-8&ctx_tim=2025-01-15T05%3A19%3A58IST&url_ver=Z39.88-2004&url_ctx_fmt=infofi/fmt:kev:mtx:ctx&rfr_id=info:sid/primo.exlibrisgroup.com:primo3-Article-arxiv_GOX&rft_val_fmt=info:ofi/fmt:kev:mtx:journal&rft.genre=article&rft.atitle=Intraoperative%202D/3D%20Image%20Registration%20via%20Differentiable%20X-ray%20Rendering&rft.au=Gopalakrishnan,%20Vivek&rft.date=2023-12-11&rft_id=info:doi/10.48550/arxiv.2312.06358&rft_dat=%3Carxiv_GOX%3E2312_06358%3C/arxiv_GOX%3E%3Curl%3E%3C/url%3E&disable_directlink=true&sfx.directlink=off&sfx.report_link=0&rft_id=info:oai/&rft_id=info:pmid/&rfr_iscdi=true