算無遺策——markerless肺結節定位法

各位新年好！肺結節是早期肺癌的常見表現，臨床診療中最常見的定位方法便是marker定位法，無論是術前CT引導下定位還是術中電磁導航、螢光現像定位，越來越多位置刁鑽的結節不斷給各種定位方法帶來挑戰。Brevity is beauty,那麼能不能僅通過影像學和立體幾何模擬來進行markerless定位呢？

A hybrid, image-based and biomechanics-based registration approach to markerless intraoperative nodule localization during video-assisted thoracoscopic surgery

Pablo Alvarez, Simon Rouzé, Michael I Miga, Yohan Payan, Jean-Louis Dillenseger, Matthieu Chabanas

Medical Image Analysis 2021 January 30, 69: 101983

The resection of small, low-dense or deep lung nodules during video-assisted thoracoscopic surgery (VATS) is surgically challenging. Nodule localization methods in clinical practice typically rely on the preoperative placement of markers, which may lead to clinical complications. We propose a markerless lung nodule localization framework for VATS based on a hybrid method combining intraoperative cone-beam CT (CBCT) imaging, free-form deformation image registration, and a poroelastic lung model with allowance for air evacuation. The difficult problem of estimating intraoperative lung deformations is decomposed into two more tractable sub-problems: (i) estimating the deformation due the change of patient pose from preoperative CT (supine) to intraoperative CBCT (lateral decubitus); and (ii) estimating the pneumothorax deformation, i.e. a collapse of the lung within the thoracic cage. We were able to demonstrate the feasibility of our localization framework with a retrospective validation study on 5 VATS clinical cases. Average initial errors in the range of 22 to 38 mm were reduced to the range of 4 to 14 mm, corresponding to an error correction in the range of 63 to 85%. To our knowledge, this is the first markerless lung deformation compensation method dedicated to VATS and validated on actual clinical data.
摘要：電視胸腔鏡下切除較小的、密度較低、位置深在的肺結節具有較大難度。臨床實踐中常用的結節定位方法是基於手術前的定位標記，由此也引發了不同程度的併發症。本文提出了一種基於多模塊的VATS下免標記肺結節定位框架，該方法結合了術中錐形束CT（CBCT）成像、無固定形式的變形圖像配準和允許空氣自由排出的多孔彈性肺模型。預估術中肺變形程度的難題被分解為兩個更易處理的子問題：（i）估計從術前CT（仰臥）到術中CBCT（側臥）患者姿勢變化而引起的形變；（ii）估計氣胸導致的形變，即胸廓內的肺萎陷。通過對5例VATS手術患者的回顧性驗證研究，研究確立了的此定位框架的可行性。應用這一方法後，定位範圍的平均初始誤差從22-38mm減少到4-14 mm，誤差範圍校正對應於63-85％。迄今為止，這是第一種專用於VATS的無需術前標記、採用肺形變補償的結節定位方法，並基於真實臨床數據進行了驗證。

Highlights
•Hybrid deformation compensation method for lung nodule localization during VATS.
•Two Cone-beam CT (CBCT) images acquired during surgery: before and after pneumothorax (lung collapse).
•Lung deformation process decomposed into (i) change of patient pose deformation and (ii) pneumothorax deformation.
•Image-based registration served to estimate boundary conditions for a biomechanical poroelastic lung model.
•Retrospective validation study on 5 VATS clinical cases.
• VATS期間肺結節定位的混合形變補償法。
• 手術期間獲取的兩張錐形束CT（CBCT）圖像：氣胸（肺塌陷）之前和之後。
• 肺部變形過程分解為（i）患者姿勢引起的形變和（ii）氣胸造成的形變。
• 基於圖像的配準有助於估計生物力學多孔肺模型的邊界。
• 對5例VATS臨床病例進行了回顧性驗證研究。

1.首先回顧一下目前各種定位方法及其優劣勢吧~！對於不同位置、不同大小、不同密度的結節，各種方法定位的成功率、併發症發生率各有不同。
2. 此文的目的就是使用術中錐形束CT和模型構建來探索實現markless肺結節定位。對這一問題的分解真是不由得讓人想起把大象裝進冰箱到底分幾步？肺結節定位大咖常說的「玄學定位法」也許就是在腦中將無數次臨床實踐中膨脹和萎陷肺的位置，進行反覆的配準糾偏。此前美國也開始了探索手術中肺萎陷後結節位置的相關研究。
3.不同於市場上常見的定位方法，此文章似乎想通過非介入的方法來進行肺結節定位。想必背後是堅實的醫學物理學巨擎，是強大的模型構建及算力。AI在肺結節定位領域鋒芒初露~！
--2020年度醫療器械註冊工作報告 來源：藥監局網站

--動脈網 

目錄 (盲猜大家從來不看英文~這次就放上中文渣翻了)
背景
組成要素
2.1. 基於強度圖像配準的肺形變補償法
2.2. 基於生物力學模型的肺形變補償法
2.3. 多模塊肺形變補償法
方法概覽 (Fig. 1-2)
肺的多孔彈性模型
CBCT圖像的預處理
步驟一：估算姿勢改變導致的形變 (Fig. 3)
6.1. 基於圖像改變的姿勢預估
  6.1.1. 脊柱的剛性定位
  6.1.2. CT圖像中肺實質的分割
  6.1.3. 初始彈性配準
  6.1.4. 精細解剖彈性配準
6.2. 全肺形變外推
  6.2.1. 有限元網格生成
  6.2.2. 位移計算
  6.2.3. 姿勢變化的有限元預估
步驟二：因氣胸所致的形變預估(Fig. 4)
7.1. 術中影像數據處理
  7.1.1. 萎陷肺表面的分割
  7.1.2. 肺門結構形變的預估
7.2. 氣胸模擬 (Fig. 5)
  7.2.1. 邊緣界定及載荷
  7.2.2. 胸廓內面模擬
  7.2.3. 膈肌面上移模擬
  7.2.4. 模型擬合肺的材料學性質 (Table 1)
7.3. 逆向擬合解決法
7.4. 結節定位
結果
8.1. 臨床數據集(Table 2; Fig. 6)
8.2. 結果: 步驟一，估算姿勢改變導致的形變 (Fig. 7-8)
8.3. 結果: 步驟二，因氣胸所致的形變預估 (Fig. 9-11; Table 3)
8.4. 控制變量法
  8.4.1. 姿勢和肺門結構改變造成的影響 (Fig. 12)
  8.4.2. 膈肌動度造成的影響 (Fig. 13)
討論
9.1. 多模塊肺形變補償法
9.2. 模型構建
9.3. 逆向擬合解決法
9.4. 膈肌動度
9.5. 走向臨床實踐：實用性和準確性
結論
附件：Qualitative results for 5 clinical cases (houlang.site/discuz/forum.php?mod=viewthread&tid=1139&extra=)
— 圖表匯總—
（限於篇幅，此處僅放上文中圖片，附表略去）

Graphical abstract

方法概覽 (Fig. 1-2)
Fig. 1. Left: preoperative CT image with the patient in supine position. Right: intraoperative CBCT images of the inflated (CBCT

Fig. 2. Overview of the proposed nodule localization framework. The process is split into two stages, Phase 1 and Phase 2, that respectively estimate the change of pose deformation then the pneumothorax deformation.

步驟一：估算姿勢改變導致的形變 (Fig. 3)
Fig. 3. Schematic diagram of the Phase 1 process to estimate the change of pose deformation. The top block illustrates the image-based registration of the preoperative CT and intraoperative CBCT
步驟二：因氣胸所致的形變預估(Fig. 4)
Fig. 4. Schematic diagram of the Phase 2 stage to estimate the pneumothorax deformation. Intraoperative images are processed to segment the surface of the deflated lung, and to compute a deformation field approximating the hilum deformation between CBCT

氣胸模擬 (Fig. 5)
Fig. 5. Schematic representation of the pneumothorax phenomenon. Left, at end of expiration the lung is in equilibrium and there is no airflow. Right, the rupture in the chest wall causes an increase of the intrapleural pressure and a decrease of the transmural pressure. The chest wall no longer pulls the surface of the lung outwards. The lung collapses due to alveoli inward recoil and gravity. The flow of air is indicated with black arrows.

臨床數據集驗證
Fig. 6. Spatial distribution of anatomical landmarks within the lung FE mesh reconstructed from the preoperative CT image.

結果: 步驟一，估算姿勢改變導致的形變
Fig. 7. TRE distributions for rigid and elastic registration between the preoperative CT and intraoperative CBCT
Fig. 8. Qualitative results of rigid and elastic registration between the preoperative CT (green) and intraoperative CBCT
結果: 步驟二，因氣胸所致的形變預估 (Fig. 9-11)
Fig. 9. TRE distributions for our complete deformation compensation framework, alongside the errors expected without deformation compensation. These latter distributions correspond to rigidly registering the preoperative CT with the CBCT
Fig. 10. Qualitative results of our deformation compensation framework for two clinical cases. Left: final deformed lung FE mesh superposed over the extracted deflated lung surface (in green). Middle: Registered landmark errors, deformed FE lung mesh and thoracic cage contact surface. Right: Initial nodule position (wireframe, black surface), ground truth nodule position (wireframe, green surface) and predicted nodule position (solid, purple surface). Results for all cases are available in the online supplementary materials. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)
Fig. 11. Qualitative results of our deformation compensation framework for two representative cases. The CT and CBCT
姿勢和肺門結構改變造成的影響 (Fig. 12)
Fig. 12. TRE distributions for three variants of the proposed lung deformation compensation method.
膈肌動度造成的影響 (Fig. 13)
Fig. 13. TRE distributions for our deformation compensation framework with and without including the upward diaphragm movement.