Staging rectal cancer: MRI compared to MDCT

PURPOSE: To audit the accuracy with which pre-operative MRI and multi-detector row CT determine the relationship between rectal tumour and the circumferential resection margin (CRM). MATERIALS AND METHODS: The MR and CT scans of 72 patients with rectal adenocarcinoma were retrieved. The relationship between tumour and the mesorectal fascia was determined by two observers, who classified appearances into three categories: no tumour within 5 mm of the mesorectal fascia; tumour within 5 mm of the mesorectal fascia ('threatened' margin); tumour at the mesorectal fascia ('involved' margin). Agreement with post-operative histopathology was assessed by Kappa statistics. RESULTS: There was poor agreement between both MRI and CT, and post-operative histology, both in all 72 patients and in the 42 who had received no pre-operative therapy or short-course radiotherapy only. Both imaging modalities had a tendency to overstage patients whose CRM was uninvolved subsequently. However, the negative predictive value for an uninvolved margin was 81.8% by MRI and 84.6% by CT. There was no patient with an involved margin by histopathology whose imaging had suggested the margin was uninvolved. CONCLUSION: Both pre-operative MRI and multi-detector row CT have high negative predictive values for a subsequently uninvolved resection margin.     

直肠腔内超声在直肠癌术前环周切缘评估中的作用

  目的  评估直肠腔内超声(endorectal ultrasound, ERUS)诊断直肠癌环周切缘(circumferential resection margin, CRM)的可行性和准确性  目的  回顾性分析2010年5月至2013年12月在北京协和医院行术前ERUS评估的直肠癌患者120例。纳入患者仅采用直肠全系膜切除术切除肿瘤, 未采用术前新辅助放化疗治疗。患者行术前ERUS检查时测量CRM, 即肿瘤的最外缘与直肠系膜筋膜的最短距离。以病理结果为金标准, 比较不同CRM诊断标准下ERUS的诊断价值。分析ERUS对不同位置、距肛缘距离、分期的CRM诊断准确性差异  结果  ERUS可以显示直肠系膜筋膜114例, 显示率为95%。采用不同探头频率, 直肠系膜筋膜显示的差异存在统计学意义(P=0.034)。以CRM ≤ 2 mm为标准时, ERUS诊断CRM的敏感性、准确性、阴性预测值最高, 分别为100%、98.2%、100%。ERUS对不同位置、距肛缘距离、病理分期病灶的诊断准确性差异不具有统计学意义(P>0.05)  结论  ERUS可以准确诊断直肠癌环周切缘, 同时具有较高的阴性预测值, 可为判断预后及制定临床治疗方案提供可靠依据。     

A magnetic resonance spectroscopy driven initialization scheme for active shape model based prostate segmentation

Segmentation of the prostate boundary on clinical images is useful in a large number of applications including calculation of prostate volume pre- and post-treatment, to detect extra-capsular spread, and for creating patient-specific anatomical models. Manual segmentation of the prostate boundary is, however, time consuming and subject to inter- and intra-reader variability. T2-weighted (T2-w) magnetic resonance (MR) structural imaging (MRI) and MR spectroscopy (MRS) have recently emerged as promising modalities for detection of prostate cancer in vivo. MRS data consists of spectral signals measuring relative metabolic concentrations, and the metavoxels near the prostate have distinct spectral signals from metavoxels outside the prostate. Active Shape Models (ASM's) have become very popular segmentation methods for biomedical imagery. However, ASMs require careful initialization and are extremely sensitive to model initialization. The primary contribution of this paper is a scheme to automatically initialize an ASM for prostate segmentation on endorectal in vivo multi-protocol MRI via automated identification of MR spectra that lie within the prostate. A replicated clustering scheme is employed to distinguish prostatic from extra-prostatic MR spectra in the midgland. The spatial locations of the prostate spectra so identified are used as the initial ROI for a 2D ASM. The midgland initializations are used to define a ROI that is then scaled in 3D to cover the base and apex of the prostate. A multi-feature ASM employing statistical texture features is then used to drive the edge detection instead of just image intensity information alone. Quantitative comparison with another recent ASM initialization method by Cosio showed that our scheme resulted in a superior average segmentation performance on a total of 388 2D MRI sections obtained from 32 3D endorectal in vivo patient studies. Initialization of a 2D ASM via our MRS-based clustering scheme resulted in an average overlap accuracy (true positive ratio) of 0.60, while the scheme of Cosio yielded a corresponding average accuracy of 0.56 over 388 2D MR image sections. During an ASM segmentation, using no initialization resulted in an overlap of 0.53, using the Cosio based methodology resulted in an overlap of 0.60, and using the MRS-based methodology resulted in an overlap of 0.67, with a paired Student's t-test indicating statistical significance to a high degree for all results. We also show that the final ASM segmentation result is highly correlated (as high as 0.90) to the initialization scheme.     

环周切缘状态对中低位直肠癌预后的影响

目的探讨环周切缘状态与中低位直肠癌患者预后的关系,并分析环周切缘(CRM)与中低位直肠癌临床病理特征的关系。方法采用大组织切片技术,对386例行全直肠系膜切除术(Total Mesorectal Excision,TME)的中低位直肠癌标本环周切缘状态进行检查,采用Kaplan-Meier法分析术后局部复发率、远处转移率和5年生存率与CRM的关系,并对临床病理特征进行分析。结果中低位直肠癌CRM阳性率为21.5%(83/386),术后局部复发率为9.0%(35/386),远处转移率为19.9%(77/386);CRM阳性的患者局部复发率为16.8%(14/83),明显高于CRM阴性的6.9%(21/303)(P=0.010);CRM≤2 cm的患者局部复发率为16.2%,而CRM2cm的患者的局部复发率为6.2%(P0.001);CRM阳性的远处转移率为34.54%(29/83),CRM阴性者为12.7%(48/303)(P=0.031);CRM阳性的5年生存率为33%,明显低于CRM阴性的78%;Kaplan-Meier生存分析显示,CRM与生存时间密切相关(log-rank,P=0.012),环周切缘状态与肿瘤直径(P=0.026)、TNM分期(P=0.021)、肿瘤距齿状线距离(P=0.009)及手术方式(P0.001)有关。结论 CRM是影响预后的一个独立因素,直肠癌患者的CRM是决定局部复发的重要因素,CRM≤2cm时局部复发率升高。     

CAN3D: Fast 3D medical image segmentation via compact context aggregation

Direct automatic segmentation of objects in 3D medical imaging, such as magnetic resonance (MR) imaging, is challenging as it often involves accurately identifying multiple individual structures with complex geometries within a large volume under investigation. Most deep learning approaches address these challenges by enhancing their learning capability through a substantial increase in trainable parameters within their models. An increased model complexity will incur high computational costs and large memory requirements unsuitable for real-time implementation on standard clinical workstations, as clinical imaging systems typically have low-end computer hardware with limited memory and CPU resources only. This paper presents a compact convolutional neural network (CAN3D) designed specifically for clinical workstations and allows the segmentation of large 3D Magnetic Resonance (MR) images in real-time. The proposed CAN3D has a shallow memory footprint to reduce the number of model parameters and computer memory required for state-of-the-art performance and maintain data integrity by directly processing large full-size 3D image input volumes with no patches required. The proposed architecture significantly reduces computational costs, especially for inference using the CPU. We also develop a novel loss function with extra shape constraints to improve segmentation accuracy for imbalanced classes in 3D MR images. Compared to state-of-the-art approaches (U-Net3D, improved U-Net3D and V-Net), CAN3D reduced the number of parameters up to two orders of magnitude and achieved much faster inference, up to 5 times when predicting with a standard commercial CPU (instead of GPU). For the open-access OAI-ZIB knee MR dataset, in comparison with manual segmentation, CAN3D achieved Dice coefficient values of (mean = 0.87 ± 0.02 and 0.85 ± 0.04) with mean surface distance errors (mean = 0.36 ± 0.32 mm and 0.29 ± 0.10 mm) for imbalanced classes such as (femoral and tibial) cartilage volumes respectively when training volume-wise under only 12G video memory. Similarly, CAN3D demonstrated high accuracy and efficiency on a pelvis 3D MR imaging dataset for prostate cancer consisting of 211 examinations with expert manual semantic labels (bladder, body, bone, rectum, prostate) now released publicly for scientific use as part of this work.     

Imaging for Predicting the Risk Factors—the Circumferential Resection Margin and Nodal Disease—of Local Recurrence in Rectal Cancer: A Meta-Analysis

The aim of the present study was to conduct a meta-analysis of English literature on the accuracy of preoperative imaging in predicting the two most important risk factors for local recurrence in rectal cancer, the circumferential resection margin (CRM) and the nodal status (N-status). Articles published between 1985 and August 2004 that report on the diagnostic accuracy of endoluminal ultrasound (EUS), computed tomography (CT), or magnetic resonance imaging (MRI) in the evaluation of lymph node involvement were included. A similar search was done for the assessment of the circumferential resection margin in rectal cancer in the period from January 1985 till January 2005. The inclusion criteria were as follows: (1) more than 20 patients with histologically proven rectal cancer were included, (2) histology was used as the gold standard, and (3) results were given in a 2 x 2 contingency table or this table could otherwise be extracted from the article by two independent readers. Based on the results summary receiver operating characteristic (ROC) curves were constructed. Only 7 articles matching inclusion criteria were found concerning the CRM. The meta-analysis shows that MRI is rather accurate in diagnosing a close or involved CRM. For nodal status 84 articles could be included. The diagnostic odds ratio of EUS is estimated at 8.83. For MRI and CT, the diagnostic odds ratio are 6.53 and 5.86, respectively. The results show that EUS is slightly, but not significantly, better than MRI or CT for identification of nodal disease. There is no significant difference between the different modalities with respect to staging nodal status. At present, MRI is the only modality that predicts the circumferential resection margin with good accuracy, making it a good tool to identify high and low risk patients. Predicting the N-status remains a problem for the radiologist for every modality, although considering the new developments in MR imaging, this may change in the near future.     

Segmentation of tongue muscles from super-resolution magnetic resonance images

Imaging and quantification of tongue anatomy is helpful in surgical planning, post-operative rehabilitation of tongue cancer patients, and studying of how humans adapt and learn new strategies for breathing, swallowing and speaking to compensate for changes in function caused by disease, medical interventions or aging. In vivo acquisition of high-resolution three-dimensional (3D) magnetic resonance (MR) images with clearly visible tongue muscles is currently not feasible because of breathing and involuntary swallowing motions that occur over lengthy imaging times. However, recent advances in image reconstruction now allow the generation of super-resolution 3D MR images from sets of orthogonal images, acquired at a high in-plane resolution and combined using super-resolution techniques. This paper presents, to the best of our knowledge, the first attempt towards automatic tongue muscle segmentation from MR images. We devised a database of ten super-resolution 3D MR images, in which the genioglossus and inferior longitudinalis tongue muscles were manually segmented and annotated with landmarks. We demonstrate the feasibility of segmenting the muscles of interest automatically by applying the landmark-based game-theoretic framework (GTF), where a landmark detector based on Haar-like features and an optimal assignment-based shape representation were integrated. The obtained segmentation results were validated against an independent manual segmentation performed by a second observer, as well as against B-splines and demons atlasing approaches. The segmentation performance resulted in mean Dice coefficients of 85.3%, 81.8%, 78.8% and 75.8% for the second observer, GTF, B-splines atlasing and demons atlasing, respectively. The obtained level of segmentation accuracy indicates that computerized tongue muscle segmentation may be used in surgical planning and treatment outcome analysis of tongue cancer patients, and in studies of normal subjects and subjects with speech and swallowing problems.     

动态增强磁共振成像对直肠癌术前新辅助治疗疗效的评价

目的观察动态增强磁共振（DCE—MRI）能否在术前判断直肠癌新辅助治疗的疗效,比较常规T2WI与DCE—MRI对直肠癌术前T、N分期的准确性。方法收集术前或术后病理确诊的直肠癌患者40例。其中行术前新辅助治疗者22例,未进行术前新辅助治疗者18例。所有患者均在GE1．5T Twinspeed HD MR扫描仪行盆腔动态增强扫描。扫描序列包括T2WI压脂、T1WI、DWI,快速三维容积T1加权脂肪抑制成像增强扫描（3DLAVA）。分析动态增强瞌线的各项参数在新辅助治疗前后的变化,观察癌灶曲线上升速率。结果①对未进行术前治疗的直肠癌分期的准确性：DCE—MRI与常规T2WI对直肠癌T分期的准确性分别为83-3％和66．7％,对环周切缘阴性判断准确性为88-2％和70．6％。T2WI及DCE—MRI均能发现直径≥2mm的淋巴结。②对新辅助治疗后直肠癌再分期的准确性：DCE—MRI对T、N再分期的准确性可达86．4％：和81.8％,对环周切缘情况判断准确性为100％。而T2WI进行T、N再分期较困难。③治疗前癌灶的曲线上升速率明显高于正常肠管;治疗后达病理完全缓解组原癌灶部略低于下段正常肠管,但是差异不明显;治疗后未达病理完全缓解组癌灶部仍高于下段正常肠管。结论DCE—MRI对术前直肠癌分期的准确性高,并能够准确判断新辅助治疗后痛周切缘情况。癌灶与参照肠管的上升速率比较有助于在术前判断疗效。     

Gastric tumorigenesis after radical resection combined with adjuvant chemotherapy for colorectal cancer: two case reports and a literature review

Radical resection with or without adjuvant chemotherapy is a common option for stage II and III colorectal cancer. Few reports exist regarding gastric tumorigenesis, including gastric cancer, gastric intraepithelial neoplasia, and gastric stromal tumor, in patients who received this protocol as the standard treatment for colorectal cancer. We present two cases of gastric tumorigenesis in patients with colorectal cancer following radical resection combined with adjuvant chemotherapy. Both patients underwent gastrectomy and D2 lymphadenectomy for their gastric tumors; neither patient developed recurrence up to 2 years after treatment. These cases indicate that patients should be monitored closely for gastric tumorigenesis after treatment for colorectal cancer. Early detection and active surgical treatment can provide satisfactory results for colorectal cancer followed by gastric tumorigenesis. Long-term follow-up and regular examinations, especially gastroscopy, are necessary to detect gastric tumorigenesis after colorectal cancer. The focus on monitoring colorectal cancer alone in colorectal cancer patients should be changed to include a broader range of cancers in addition to precancers and other tumors, such as gastric stromal tumor.     

Suggestive annotation of brain MR images with gradient-guided sampling

Machine learning has been widely adopted for medical image analysis in recent years given its promising performance in image segmentation and classification tasks. The success of machine learning, in particular supervised learning, depends on the availability of manually annotated datasets. For medical imaging applications, such annotated datasets are not easy to acquire, it takes a substantial amount of time and resource to curate an annotated medical image set. In this paper, we propose an efficient annotation framework for brain MR images that can suggest informative sample images for human experts to annotate. We evaluate the framework on two different brain image analysis tasks, namely brain tumour segmentation and whole brain segmentation. Experiments show that for brain tumour segmentation task on the BraTS 2019 dataset, training a segmentation model with only 7% suggestively annotated image samples can achieve a performance comparable to that of training on the full dataset. For whole brain segmentation on the MALC dataset, training with 42% suggestively annotated image samples can achieve a comparable performance to training on the full dataset. The proposed framework demonstrates a promising way to save manual annotation cost and improve data efficiency in medical imaging applications.     

Structured patch model for a unified automatic and interactive segmentation framework

We present a novel interactive segmentation framework incorporating a priori knowledge learned from training data. The knowledge is learned as a structured patch model (StPM) comprising sets of corresponding local patch priors and their pairwise spatial distribution statistics which represent the local shape and appearance along its boundary and the global shape structure, respectively. When successive user annotations are given, the StPM is appropriately adjusted in the target image and used together with the annotations to guide the segmentation. The StPM reduces the dependency on the placement and quantity of user annotations with little increase in complexity since the time-consuming StPM construction is performed offline. Furthermore, a seamless learning system can be established by directly adding the patch priors and the pairwise statistics of segmentation results to the StPM. The proposed method was evaluated on three datasets, respectively, of 2D chest CT, 3D knee MR, and 3D brain MR. The experimental results demonstrate that within an equal amount of time, the proposed interactive segmentation framework outperforms recent state-of-the-art methods in terms of accuracy, while it requires significantly less computing and editing time to obtain results with comparable accuracy.     

影响低位直肠癌手术保肛相关因素的研究

目的分析影响低位直肠癌手术保肛的相关因素。方法将341例低位直肠癌患者按是否施行保肛手术而分为APR组和SP组，对两组病例临床资料（包括肿瘤下缘距肛缘的距离、患者年龄、性别、BMI）及病理资料（包括肿瘤周径、肿瘤位置、肿瘤大体分型、分化程度和Dukes分期）进行比较分析。结果APR组肿瘤下缘距肛缘距离显著小于SP组，肿瘤周径小于1／2周、1／2～3／4周、大于3／4周及全周四者保肛率依次下降（P〈0．01），对于肿瘤周径小于1／2周者，肿瘤主体位于后壁、侧壁及前壁者三者保肛率依次下降，但无明显统计学意义（P〉0．05）。肿瘤大体分型中，肿块型、溃疡型及浸润型三者保肛率依次下降（P〈0．05），肿瘤病理呈高、中、低分化癌者三者保肛率逐渐下降（P〈0．01），低位直肠癌Dukes A、B、C、D期四者保肛率呈逐渐下降趋势。青年人、男性、BMI≥23的患者其保肛率显著低于中老年人、女性及BMI〈23的患者（P〈0．05）。结论肿瘤下缘距齿状线或肛缘的距离是低位直肠癌手术保肛最主要的影响因素，肿瘤周径、肿瘤病理分期、肿瘤主体位置、肿瘤大体分型、肿瘤分化程度及患者的性别及BMI等均可对低位直肠癌手术保肛产生影响。     

Uncertainty and interpretability in convolutional neural networks for semantic segmentation of colorectal polyps

Colorectal polyps are known to be potential precursors to colorectal cancer, which is one of the leading causes of cancer-related deaths on a global scale. Early detection and prevention of colorectal cancer is primarily enabled through manual screenings, where the intestines of a patient is visually examined. Such a procedure can be challenging and exhausting for the person performing the screening. This has resulted in numerous studies on designing automatic systems aimed at supporting physicians during the examination. Recently, such automatic systems have seen a significant improvement as a result of an increasing amount of publicly available colorectal imagery and advances in deep learning research for object image recognition. Specifically, decision support systems based on Convolutional Neural Networks (CNNs) have demonstrated state-of-the-art performance on both detection and segmentation of colorectal polyps. However, CNN-based models need to not only be precise in order to be helpful in a medical context. In addition, interpretability and uncertainty in predictions must be well understood. In this paper, we develop and evaluate recent advances in uncertainty estimation and model interpretability in the context of semantic segmentation of polyps from colonoscopy images. Furthermore, we propose a novel method for estimating the uncertainty associated with important features in the input and demonstrate how interpretability and uncertainty can be modeled in DSSs for semantic segmentation of colorectal polyps. Results indicate that deep models are utilizing the shape and edge information of polyps to make their prediction. Moreover, inaccurate predictions show a higher degree of uncertainty compared to precise predictions.     

Registration of 3D CT angiography and cardiac MR images in coronary artery disease patients

A method for the registration of 3D cardiac CT angiography (CTA) and magnetic resonance (MR) data sets based on their myocardial epicardial surfaces is introduced. The approach relies on temporally registered data sets obtained based on the electrocardiogram recorded during the CTA acquisition and the timing characteristics of the MR acquisition. The myocardial epicardial surfaces were identified in the reformatted CTA and MR data sets using a 3D semi-automated segmentation algorithm. This algorithm was implemented, evaluated on clinical data, and compared to a set of manual outlines during the course of this study. The registration of the CTA and MR data sets was based on the iterative closest point algorithm, which minimizes the distance between the surfaces defined by the epicardial outlines in each data set. The proposed technique was applied to data obtained from 11 patients with coronary artery disease. The CTA data was reformatted based on the registration results and the location of the MR imaging planes. The resulting CTA-MR image pairs were evaluated qualitatively by two experts, who graded the majority of the cases as either excellent or acceptable (11 of 11 cases for one reader, and 9 of 11 for the other). The results were evaluated quantitatively based on the distance between the registered epicardial surfaces. The quantitative measures indicated that the registered surfaces were within two pixels of one another (on average). The registration results were used to generate combined 3D renderings of information extracted from both data sets for visualization purposes.     

MRI观察壁外血管侵犯评估直肠癌新辅助治疗后根治性手术患者预后

目的 探讨MRI观察直肠壁外血管侵犯（mrEMVI）以评估直肠癌新辅助治疗（NAT）及根治术后患者预后的价值。方法 回顾性分析61例于NAT后接受根治性手术的直肠癌患者，均于NAT前接受3.0T高分辨率MR检查，分为mrEMVI阳性组与阴性组，对比2组相关资料；术后随访，统计3年无瘤生存率（DFS），采用COX单因素及多因素回归分析观察DFS影响因素。结果 mrEMVI阳性组31例，阴性组30例，MRI所示组间T分期、浸润深度、肿瘤位置及环周切缘阳性率差异均有统计学意义（P均<0.05）。中位随访时间29个月[95% CI（1.1，50.0）]，mrEMVI阴性患者3年DFS为82.5%，明显高于mrEMVI阳性者（48.8%，P=0.013）。mrEMVI及病理分级可预测3年DFS（P均<0.05）。结论 mrEMVI可用于评估直肠癌NAT及根治术后患者预后。     

KCB-Net: A 3D knee cartilage and bone segmentation network via sparse annotation

Knee cartilage and bone segmentation is critical for physicians to analyze and diagnose articular damage and knee osteoarthritis (OA). Deep learning (DL) methods for medical image segmentation have largely outperformed traditional methods, but they often need large amounts of annotated data for model training, which is very costly and time-consuming for medical experts, especially on 3D images. In this paper, we report a new knee cartilage and bone segmentation framework, KCB-Net, for 3D MR images based on sparse annotation. KCB-Net selects a small subset of slices from 3D images for annotation, and seeks to bridge the performance gap between sparse annotation and full annotation. Specifically, it first identifies a subset of the most effective and representative slices with an unsupervised scheme; it then trains an ensemble model using the annotated slices; next, it self-trains the model using 3D images containing pseudo-labels generated by the ensemble method and improved by a bi-directional hierarchical earth mover’s distance (bi-HEMD) algorithm; finally, it fine-tunes the segmentation results using the primal–dual Internal Point Method (IPM). Experiments on four 3D MR knee joint datasets (the SKI10 dataset, OAI ZIB dataset, Iowa dataset, and iMorphics dataset) show that our new framework outperforms state-of-the-art methods on full annotation, and yields high quality results for small annotation ratios even as low as 10%.     

The postoperative brain tumour stem cell (BTSC) niche and cancer recurrence

Currently, surgical resection is one of only a few options for treating brain cancer. Unfortunately, postoperative tumour recurrence remains almost inevitable despite additional radiation or chemotherapy treatment following resection. Clinical observations and a growing body of experimental evidence have led to speculation that there is a population of persistent brain tumour stem cells (BTSCs) — or brain tumour initiating cells — that are difficult to completely remove surgically. Furthermore, residual BTSCs following surgery may actually be more resistant to subsequent radiation and/or chemotherapies. It remains to be determined if brain surgeries render the postoperative tissue microenvironment more favourable for the survival and growth of BTSCs, and therefore the recurrence of brain tumours. We hypothesise that BTSC-based tumour recurrence may develop within a specific niche of the aberrant tumour microenvironment. Even when the gross appearance of the primary tumour seems confined, BTSCs (albeit accounting only for a small population of tumour cells) may microscopically enter the stroma, hampering curative surgeries. This article discusses the theory that surgical resection of brain tumours generates niches recruiting BTSCs to the surgical wounds, stimulating the proliferation and invasiveness of BTSCs, and leading to tumour recurrence. Postoperative brains are marked with active wound repair in peritumoural margins, which is likely to be accompanied by increased inflammatory paracrine production, angiogenesis and reactive astrogliosis. The postoperative BTSC niche concept is consistent with the observation that brain tumour recurrence usually occurs in tissues that are proximal to the resection margin. In this article, we intend to reflect recent advances that may lead to novel strategies to eliminate postoperative brain tumour recurrence.     

On the usability of synthetic data for improving the robustness of deep learning-based segmentation of cardiac magnetic resonance images

Deep learning-based segmentation methods provide an effective and automated way for assessing the structure and function of the heart in cardiac magnetic resonance (CMR) images. However, despite their state-of-the-art performance on images acquired from the same source (same scanner or scanner vendor) as images used during training, their performance degrades significantly on images coming from different domains. A straightforward approach to tackle this issue consists of acquiring large quantities of multi-site and multi-vendor data, which is practically infeasible. Generative adversarial networks (GANs) for image synthesis present a promising solution for tackling data limitations in medical imaging and addressing the generalization capability of segmentation models. In this work, we explore the usability of synthesized short-axis CMR images generated using a segmentation-informed conditional GAN, to improve the robustness of heart cavity segmentation models in a variety of different settings. The GAN is trained on paired real images and corresponding segmentation maps belonging to both the heart and the surrounding tissue, reinforcing the synthesis of semantically-consistent and realistic images. First, we evaluate the segmentation performance of a model trained solely with synthetic data and show that it only slightly underperforms compared to the baseline trained with real data. By further combining real with synthetic data during training, we observe a substantial improvement in segmentation performance (up to 4% and 40% in terms of Dice score and Hausdorff distance) across multiple data-sets collected from various sites and scanner. This is additionally demonstrated across state-of-the-art 2D and 3D segmentation networks, whereby the obtained results demonstrate the potential of the proposed method in tackling the presence of the domain shift in medical data. Finally, we thoroughly analyze the quality of synthetic data and its ability to replace real MR images during training, as well as provide an insight into important aspects of utilizing synthetic images for segmentation.     

术前MRI对直肠癌分期和侧切缘的预测价值

目的：评估术前MRI预测直肠癌T、N分期和侧切缘（Circumferential resection margin,CRM）受累的准确程度。方法：术前肠镜活检病理证实直肠癌28例,MRI检查前经肛门插入三腔管气囊（内注水或气适量）,采用Siemens Avanto 1.5T磁共振系统,用MRI评估肿瘤T分期﹑系膜淋巴结转移N分期和CRM受累,所有病例均行全直肠系膜切除术TME,对照术前MRI分期和术后病理结果,评估MRI能否准确预测直肠癌T、N分期及CRM受累。结果：MRI正确T分期24例,错误4例,其中3例T1～T2期高估为T3期,1例T3期低估为T1～T2期,T分期的总准确率为85.7%（24/28）,其中T1～T2期的预测准确率为66.7%（6/9）,T3期的准确率为92.3%（12/13）,T4期的准确率为100%（6/6）,MRI可以对T分期进行准确预测（Kappa值为0.773,P<0.001）。MRI对直肠系膜淋巴结正确分期22例,错误分期6例,其中高估4例,低估2例,准确率为78.6%（22/28）,灵敏度为89.5%（17/19）,特异度为55.6%（5/9）,MR对系膜淋巴结转移可做出一般性预测（Kappa值为0.478,P=0.01）。MRI正确预测CRM 26例,错误2例,其中高估1例,低估1例,准确率为92.9%（26/28）,灵敏度为85.7%（6/7）,特异度为95.2%（20/21）。MRI可以准确预测CRM是否受累（Kappa值为0.81,P<0.001）。结论：使用三腔管气囊进行术前MRI检查能准确预测直肠癌T分期,并可靠地预测CRM是否受累,对N分期只能一般性预测。