2017年国际脑损伤与肿瘤研讨会论文精选：深度探讨脑瘤、多发性硬化与中风

医学图像处理

需积分: 9 54 浏览量更新于2024-07-17 1 收藏 99.86MB PDF 举报

身份认证购VIP最低享 7 折!

领优惠券(最高得80元）

《Brainlesion: Glioma, Multiple Sclerosis, Stroke and Traumatic Brain Injuries 3》是一本收录了第三国际MICCAI Brainlesion Workshop（BrainLes 2017）精选论文的书籍，该会议于2017年9月在加拿大魁北克市举行的Medical Image Computing for Computer Assisted Intervention (MICCAI) 大会上联合举行。书中包含了40篇经过精心筛选的文章，这些论文是从46份提交作品中挑选出来的，内容涵盖了脑部病变图像分析、脑肿瘤图像分割以及缺血性中风病变图像分割等多个主题。本书由Alessandro Crimi、Spyridon Bakas和Hugo Kuijf、Bjoern Menze共同编辑，他们的专业背景在医学图像处理领域具有深厚的影响力。这些论文反映了当时在识别和分析脑瘤（如胶质瘤）、多发性硬化症、中风和创伤性脑损伤等神经退行性疾病及其相关影像特征的最新研究进展。会议特别关注了多模态脑肿瘤分割挑战（BraTS）和白质高信号影（WMH）的分割技术，这些都是神经影像学中的重要研究方向，有助于提高临床诊断的精确性和早期干预的可能性。编者们，包括David Hutchison、Takeo Kanade等知名学者，集合了全球顶尖的研究力量，他们的合作展示了在利用计算机辅助方法处理复杂的脑部疾病成像数据方面的前沿成果。书中探讨的算法和技术不仅限于理论层面，还可能涉及深度学习、机器学习、图像处理算法的创新应用，以及如何将这些技术转化为实际临床工具。通过对这本书的学习，读者能够深入了解当前在处理脑部病变时所使用的先进方法，例如通过深度卷积神经网络进行精准的病灶定位，或者通过多模态融合来增强对脑结构的理解。此外，书中可能还涵盖了如何处理和解释不同模态数据（如MRI、CT或PET），以及如何解决由于病患个体差异带来的挑战等问题。《Brainlesion: Glioma, Multiple Sclerosis, Stroke and Traumatic Brain Injuries 3》是医学图像处理和计算机辅助诊断领域的重要参考资料，对于研究人员、医生以及希望了解和应用这些先进技术的学生来说，都是不可或缺的一份学习资料。通过阅读和研究这本书，读者不仅可以扩展专业知识，还能了解到未来该领域的发展趋势和潜在的研究方向。

资源详情

资源推荐

4I.Oguzetal.

new task. In medical image segmentation, the Dice [6] and Jaccard [11] overlap

ratios are the most popular evaluation measures [16,22]. The mean and maxi-

mum surface distances between 3D objects are also commonly used to evaluate

segmentation algorithms [9], as are distances between segmentation results and

manual landmarks. However, such measures are based on the implicit assump-

tion that the number of segmentation targets is known a priori, e.g., whether it

is one liver, two hippocampi, or ﬁve lung lobes. This is in contrast to a diﬀer-

ent but equally important class of segmentation tasks, such as the segmentation

of white matter lesions or lung nodules, where the number of target objects

can vary from zero to hundreds or more. In these scenarios, the object detection

and segmentation tasks become intertwined and as such, performance evaluation

needs to take both aspects into consideration. Although, it has been customary

to use the image-wide Dice overlap in these combined detection and segmen-

tation problems, this is an oversimpliﬁcation and can often hide diﬀerences in

algorithm behaviors (e.g., Fig. 1). Another popular criterion in these types of

applications is the number of segmented objects, which attempts to evaluate the

detection success. However, the object count alone is an ambiguous metric as it

reﬂects not only the false positive and false negative detections but also larger

objects that may be erroneously split into multiple smaller ones, or multiple

small objects erroneously merged into a single, larger object. Furthermore, these

commonly used metrics fail to relate the size of the object to the ﬁnal score,

which can be important: for many applications, an algorithm that misses large

objects is considered less clinically relevant than an algorithm that misses small

objects.

A prime example of an application domain with a variable number of objects

is multiple sclerosis (MS) lesion segmentation from MRI scans of the brain. White

matter lesions are the hallmark of MS and their segmentation and quantiﬁca-

tion are critical for clinical purposes. Many approaches to MS lesion segmen-

tation have been proposed: artiﬁcial [10] and convolutional neural networks [1];

Bayesian models [7]; Gaussian mixture models [23]; graph cuts [8]; and ran-

dom forests [12]. This ﬁeld of research remains very active and several grand

challenges (MICCAI 2008 [19], ISBI 2015 [2], MICCAI 2016 [13]) have been

organized in recent years. The evaluation of new algorithms often relies on Dice

and Jaccard overlaps and lesion counts, which limits our ability to fully assess

other characteristics in performance diﬀerence.

We propose a new set of evaluation techniques to compare segmentations

with a variable number of target objects, including a classiﬁcation of segmenta-

tion results and statistics at object and category levels. We illustrate these tech-

niques in an MS lesion segmentation study comparing two algorithms: Lesion-

TOADS [17] which is an unsupervised clustering algorithm with topological con-

straints; and OASIS [21] is a supervised classiﬁer based on multi-modal logistic

regression.

6I.Oguzetal.

2.2 Object Correspondence Identiﬁcation and Classiﬁcation

Given a pair of segmentations S

and S

∈{0, 1}

for a subject i, e.g., an

automated and a manual segmentation, respectively, we begin by identifying

the (6-) connected components in each segmentation as individual objects (i.e.,

lesions in our application). We name these two sets of objects C

and C

. Then,

for each object in C

, we identify its corresponding objects by determining all

objects in C

(if any) that it overlaps with. Formally, for an object O

ij1

∈ C

the set of matching objects is m(O

ij1

)={O

ik2

, ∀k : O

ij1

∩ O

ik2

= ∅}. Note that

this means 0, 1, or multiple corresponding objects are possible. Similarly, for

each object O

ij2

∈ C

, we identify its set of corresponding objects, m(O

ij2

), by

determining all objects in C

(if any) that it overlaps with.

Following correspondence identiﬁcation, it is possible to determine the match

conﬁguration for each object by considering the number of forward and backward

correspondences. For example, if an object O

ij1

∈ C

corresponds to a single

object O

ik2

∈ C

,andO

ik2

corresponds to a single object (i.e., O

ij1

), then we

have a 1-1 match. In contrast, if an object O

ij1

∈ C

corresponds to multiple

objects O

i12

-O

iN2

, and each of the O

i·2

correspond only to O

ij1

, then we have a

1-N match. Following the nomenclature of [15], we have the following categories:

– Correct detection: 1-1 match.

– False alarm: 1-0 match. An object exists in C

that does not exist in C

i.e., a false positive if C

is the truth.

– Merge: 1-N match. Multiple objects in C

are merged into a single object

in C

– Split: N-1 match. A single object in C

is split into multiple objects in C

– Split-merge: N-N match. The conditions for both merge and split are satis-

ﬁed.

– Detection failure: 0-1 match. An object exists in C

that does not exist in

, i.e., a false negative if C

is the truth.

We note that while the results presented in this paper focus on the compari-

son of an automated segmentation S

and a manual segmentation S

, the same

idea for classiﬁcation can also be applied to the comparison of two manual seg-

mentations to assess intra- or inter-rater variability, as well as to segmentations

from diﬀerent timepoints in a longitudinal study to assess the disease course.

2.3 Lesion Segmentation Evaluation

We propose a battery of statistics to compare two binary segmentations for a

subject i, S

and S

, which will be illustrated in Sect. 3.

We begin by reporting the classical image-wide overlap measures. In partic-

ular, for each subject i, we report the Dice overlap (DSC

∩S

|+|S

)[6],

the Jaccard overlap (2

∩S

∪S

)[11], target overlap (

∩S

), the false negative

error (

−S

) and the false positive error (

−S

). We also report the num-

ber of connected objects (|C

|), which is another popular evaluation measure.

Dice Overlap Measures for Multiple Objects 7

Fig. 2. For both LesionTOADS and OASIS, we show the (a) Dice overlap by lesion

class and the (b) lesion count by class.

Next, we classify the object segmentation as described in Sect. 2.2 and report

the number of objects n

ik1

and n

ik2

, deﬁned by S

and S

, in each cat-

egory k. We further report the average per-object Dice overlap,

DSC

ik1

−1

ik1



ik1

j=1

DSC

ijk

, for each category k, where DSC

ijk

is the Dice of an object j,

deﬁned as the Dice overlap between the segmentation of the j

object of type

k in S

1, O

ijk1

, and the matching set m(O

ijk1

Next, we analyze the mean per-object Dice overlap as a function of true object

size, i.e., f (s)=E(DSC

ijk

||O

ijk

| = s), where |·| denotes the volume of an object

and E is the expectation operator. This is done both globally for all objects, as

well as separately for each category of matches (which we denote by f

(s)), using

locally weighted scatterplot smoothing (LOESS) [3]. We use scatterplots of Dice

vs. true object size to visualize this data. To estimate 95% conﬁdence bands

for the LOESS scatterplots, we use a nonparametric bootstrap. We resample by

subject to respect the nested object-within-subject correlation structure. That is,

for each b ∈ [1, 10000], we resample subjects i ∈ [1,n] with replacement to form

a bootstrapped sample of pairs of segmentations (S

(b)

),...,(S

(b)

)and

re-estimate f(s), and f

(s) and denote these estimates by

(b)

(s), and

(b)

(s).

We then calculate the pointwise 2.5% and 97.5% quantiles of

(b)

and

(b)

estimate the lower and upper limits of conﬁdence bands. For the false alarm

and correct detection classes, we use histograms of object size (since the Dice

is always 0 for these classes). Additionally, for these two classes, we report the

spatial distribution of the occurrences by registering all images into a common

atlas space to construct a spatial occurrence frequency map.

3 Results

Figure 1 shows a summary of the whole-image overlap measures. LesionTOADS

and OASIS have comparable Dice and Jaccard measures, which is important

since these are two of the most commonly used measures for comparing segmen-

tation algorithms. LesionTOADS has a smaller false negative ratio but a higher

false positive ratio than OASIS. The number of distinct lesions as segmented by

the expert manual rater was 2461; OASIS detected 1340 distinct lesions whereas

LesionTOADS detected 2810. LesionTOADS reports more lesions than OASIS,

8I.Oguzetal.

but it is uncertain using these measures whether this is driven by a higher suc-

cessful detection rate, more false alarms, more split lesions, or other reasons.

Figure 2(a) shows the per-lesion Dice overlap, summarized per class. Note

that the Dice for the detection failures and false alarms is 0 by construction. The

LesionTOADS algorithm has a higher Dice than OASIS in every category, which

is rather surprising given that the whole-image Dice measures are nearly equal

between the two methods (see Fig. 1). Figure 2(b) shows the number of lesions

in each of the classes described in Sect. 2. Compared to overall lesion counts,

this analysis provides further insight into the algorithm behavior: compared to

OASIS, LesionTOADS has a larger number of correctly detected lesions (good),

and fewer detection failures (good), but also more merges (bad), and many more

false alarms (bad). As such, it is diﬃcult to declare an overall “winner” but each

algorithm is “winning” for diﬀerent classes of lesions.

Figure 3(a) shows the per-lesion Dice overlap as a function of true lesion size.

On average, LesionTOADS seems to perform better than OASIS for small and

large lesions, whereas OASIS performs better for the more prevalent medium-

sized lesions. It is also interesting that in this medium-size range, OASIS appears

to have a tighter distribution of Dice scores whereas LesionTOADS performs

either very well (Dice > 0.8) or very poorly (Dice < 0.2). Figures 3(c) and (d)

provide additional insight by breaking down this data into individual classes.

Figure 4 takes this analysis one step further by directly comparing the algo-

rithms’ behaviors in each class. Figure 4(a) shows this comparison for the correctly

detected lesions. We note that the average performance of the LesionTOADS algo-

rithm increases steeply with size in this class, indicating the algorithm is highly

accurate for all but the smallest lesions (which are notoriously diﬃcult to seg-

ment correctly), for those lesions that it manages to detect correctly. In contrast,

OASIS performance improves more slowly with lesion size. Figure 4(b) compares

the two methods for merged lesions; while the performance of the two algorithms

are roughly comparable and both improve with lesion size, there are overall fewer

merged lesions for OASIS, which is desirable. Figures 4(c) and (d) provide the

same comparison for split and split-merge classes, respectively.

For false alarms and detection failures, instead of scatterplots, we present

spatial distribution maps and size histograms of lesions. Figure 5 shows the detec-

tion failures for the two algorithms. It is interesting that the distribution of these

failures are remarkably similar between the two algorithms for smaller lesions,

suggesting many of these smaller lesions may be generally diﬃcult to detect. The

spatial distributions of these detection failures concentrate on the septum area

for both methods. OASIS has an additional hotspot for detection failures near

the temporal horn of the ventricles.

Figure 6 compares the false alarms for the two algorithms. LesionTOADS

appears to generate hardly any small false positive lesions, but many medium-to-

large false positive lesions. This rather surprising ﬁnding explains the counterin-

tuitive result that while LesionTOADS reports better Dice overlap for each lesion

category (Fig. 2), the whole-image Dice scores are nearly identical between the

two methods (see Fig. 1). We note that the large number of false alarms reported

剩余523页未读，继续阅读

wang1062807258

粉丝: 13
资源: 272

2017年国际脑损伤与肿瘤研讨会论文精选：深度探讨脑瘤、多发性硬化与中风

A Survey and Analysis on Automated Glioma Brain Tumor Seg

脑肿瘤分类数据集7023张4类别.7z

Gene level pathogenesis of brain glioma

I=imread('D:\srt\Brain Tumor MRI Dataset\Testing\glioma\Te-gl_0010.gif');这句话错在哪里了

有哪些有详细介绍CCK8测定细胞活性原理的文献

基于kmeans聚类算法的胶质瘤分级

细胞增殖对细胞活力测定影响的参考文献】

细胞增殖对细胞活力测定影响的参考文献

影像组学全流程 R语言

R语言读取CGGA临床数据

EGFRvIII、HER2、IL-13Rα2翻译成中文及其含义

EGFRvIII、HER2、IL-13Rα2

IL-13Rα2 CAR是什么意思

基于微信小程序校园订餐的+ssm演示+源代码+开题报告+演示视频.zip

java项目实战练习.zip

Java项目-基于微信小程序的基于校园作业反馈的家校联系微信小程序（包括源码，数据库，教程）.zip

java抗疫物质管理系统设计和实现论文.docx

MPC与LQR最优控制的区别2

最新资源