Simon K. Warfield

Simultaneous truth and performance level estimation (STAPLE): an algorithm for the validation of image segmentation

Characterizing the performance of image segmentation approaches has been a persistent challenge. Performance analysis is important since segmentation algorithms often have limited accuracy and precision. Interactive drawing of the desired segmentation by human raters has often been the only acceptable approach, and yet suffers from intra-rater and inter-rater variability. Automated algorithms have been sought in order to remove the variability introduced by raters, but such algorithms must be assessed to ensure they are suitable for the task. The performance of raters (human or algorithmic) generating segmentations of medical images has been difficult to quantify because of the difficulty of obtaining or estimating a known true segmentation for clinical data. Although physical and digital phantoms can be constructed for which ground truth is known or readily estimated, such phantoms do not fully reflect clinical images due to the difficulty of constructing phantoms which reproduce the full range of imaging characteristics and normal and pathological anatomical variability observed in clinical data. Comparison to a collection of segmentations by raters is an attractive alternative since it can be carried out directly on the relevant clinical imaging data. However, the most appropriate measure or set of measures with which to compare such segmentations has not been clarified and several measures are used in practice. We present here an expectation-maximization algorithm for simultaneous truth and performance level estimation (STAPLE). The algorithm considers a collection of segmentations and computes a probabilistic estimate of the true segmentation and a measure of the performance level represented by each segmentation. The source of each segmentation in the collection may be an appropriately trained human rater or raters, or may be an automated segmentation algorithm. The probabilistic estimate of the true segmentation is formed by estimating an optimal combination of the segmentations, weighting each segmentation depending upon the estimated performance level, and incorporating a prior model for the spatial distribution of structures being segmented as well as spatial homogeneity constraints. STAPLE is straightforward to apply to clinical imaging data, it readily enables assessment of the performance of an automated image segmentation algorithm, and enables direct comparison of human rater and algorithm performance.

Abnormal Cerebral Structure Is Present at Term in Premature Infants

Background. Long-term studies of the outcome of very prematurely born infants have clearly documented that the majority of such infants have significant motor, cognitive, and behavioral deficits. However, there is a limited understanding of the nature of the cerebral abnormality underlying these adverse neurologic outcomes. Aim. The overall aim of this study was to define quantitatively the alterations in cerebral tissue volumes at term equivalent in a large longitudinal cohort study of very low birth weight premature infants in comparison to term-born infants by using advanced volumetric 3-dimensional magnetic resonance imaging (MRI) techniques. We also aimed to define any relationship of such perinatal lesions as white matter (WM) injury or other potentially adverse factors to the quantitative structural alterations. Additionally, we wished to identify the relationship of the structural alterations to short-term neurodevelopmental outcome. Methods. From November 1998 to December 2000, 119 consecutive premature infants admitted to the neonatal intensive care units at Christchurch Women’s Hospital (Christchurch, New Zealand) and the Royal Women’s Hospital (Melbourne, Australia) were recruited (88% of eligible) after informed parental consent to undergo an MRI scan at term equivalent. Twenty-one term-born infants across both sites were recruited also. Postacquisition advanced 3-dimensional tissue segmentation with 3-dimensional reconstruction was undertaken to estimate volumes of cerebral tissues: gray matter (GM; cortical and deep nuclear structures), WM (myelinated and unmyelinated), and cerebrospinal fluid (CSF). Results. In comparison to the term-born infants, the premature infants at term demonstrated prominent reductions in cerebral cortical GM volume (premature infants [mean ± SD]: 178 ± 41 mL; term infants: 227 ± 26 mL) and in deep nuclear GM volume (premature infants: 10.8 ± 4.1 mL; term infants: 13.8 ± 5.2 mL) and an increase in CSF volume (premature infants: 45.6 ± 22.1 mL; term infants: 28.9 ± 16 mL). The major predictors of altered cerebral volumes were gestational age at birth and the presence of cerebral WM injury. Infants with significantly reduced cortical GM and deep nuclear GM volumes and increased CSF volume volumes exhibited moderate to severe neurodevelopmental disability at 1 year of age. Conclusions. This MRI study of prematurely born infants further defines the nature of quantitative cerebral structural abnormalities present as early as term equivalent. The abnormalities particularly involve cerebral neuronal regions including both cortex and deep nuclear structures. The pattern of cerebral alterations is related most significantly to the degree of immaturity at birth and to concomitant WM injury. The alterations are followed by abnormal short-term neurodevelopmental outcome.

Statistical validation of image segmentation quality based on a spatial overlap index.

RATIONALE AND OBJECTIVES To examine a statistical validation method based on the spatial overlap between two sets of segmentations of the same anatomy. MATERIALS AND METHODS The Dice similarity coefficient (DSC) was used as a statistical validation metric to evaluate the performance of both the reproducibility of manual segmentations and the spatial overlap accuracy of automated probabilistic fractional segmentation of MR images, illustrated on two clinical examples. Example 1: 10 consecutive cases of prostate brachytherapy patients underwent both preoperative 1.5T and intraoperative 0.5T MR imaging. For each case, 5 repeated manual segmentations of the prostate peripheral zone were performed separately on preoperative and on intraoperative images. Example 2: A semi-automated probabilistic fractional segmentation algorithm was applied to MR imaging of 9 cases with 3 types of brain tumors. DSC values were computed and logit-transformed values were compared in the mean with the analysis of variance (ANOVA). RESULTS Example 1: The mean DSCs of 0.883 (range, 0.876-0.893) with 1.5T preoperative MRI and 0.838 (range, 0.819-0.852) with 0.5T intraoperative MRI (P < .001) were within and at the margin of the range of good reproducibility, respectively. Example 2: Wide ranges of DSC were observed in brain tumor segmentations: Meningiomas (0.519-0.893), astrocytomas (0.487-0.972), and other mixed gliomas (0.490-0.899). CONCLUSION The DSC value is a simple and useful summary measure of spatial overlap, which can be applied to studies of reproducibility and accuracy in image segmentation. We observed generally satisfactory but variable validation results in two clinical applications. This metric may be adapted for similar validation tasks.

Improved watershed transform for medical image segmentation using prior information

The watershed transform has interesting properties that make it useful for many different image segmentation applications: it is simple and intuitive, can be parallelized, and always produces a complete division of the image. However, when applied to medical image analysis, it has important drawbacks (oversegmentation, sensitivity to noise, poor detection of thin or low signal to noise ratio structures). We present an improvement to the watershed transform that enables the introduction of prior information in its calculation. We propose to introduce this information via the use of a previous probability calculation. Furthermore, we introduce a method to combine the watershed transform and atlas registration, through the use of markers. We have applied our new algorithm to two challenging applications: knee cartilage and gray matter/white matter segmentation in MR images. Numerical validation of the results is provided, demonstrating the strength of the algorithm for medical image segmentation.

Periventricular white matter injury in the premature infant is followed by reduced cerebral cortical gray matter volume at term

Periventricular white matter injury, that is, periventricular leukomalacia (PVL), the dominant form of brain injury in the premature infant, is the major neuropathological substrate associated with the motor and cognitive deficits observed later in such infants. The nature of the relationship of this lesion to the subsequent cognitive deficits is unclear, but such deficits raise the possibility of cerebral cortical neuronal dysfunction. Although cortical neuronal necrosis is not a prominent feature of brain injury in premature infants, the possibility of a deleterious effect of PVL on subsequent cerebral cortical development has not been investigated. An advanced quantitative volumetric three‐dimensional magnetic resonance imaging technique was used to measure brain tissue volumes at term in premature infants with earlier ultrasonographic and magnetic resonance imaging evidence of PVL (mean gestational age at birth, 28.7 ± 2.0 weeks; n = 10), in premature infants with normal imaging studies (mean gestational age at birth, 29.0 ± 2.1 weeks; n = 10), and in control term infants (n = 14). Premature infants with PVL had a marked reduction in cerebral cortical gray matter at term compared with either premature infants without PVL or normal term infants (mean ± SD: PVL, 157.5 ± 41.5 ml; no PVL, 211.7 ± 25.4 ml; normal term, 218.8 ± 21.3 ml). As expected, a reduction in the volume of total brain myelinated white matter was also noted (mean ± SD: PVL, 14.5 ± 4.6 ml; no PVL, 23.1 ± 6.9 ml; normal term, 27.6 ± 10.3 ml). An apparent compensatory increase in total cerebrospinal fluid volume also was found (mean ± SD: PVL, 64.5 ± 15.2 ml; no PVL, 52.0 ± 24.1 ml; normal term, 32.9 ± 13.5 ml). PVL in the premature infant is shown for the first time to be followed by impaired cerebral cortical development. These findings may provide insight into the anatomical correlate for the intellectual deficits associated with PVL in the premature infant.

Adaptive, template moderated, spatially varying statistical classification.

A novel image segmentation algorithm was developed to allow the automatic segmentation of both normal and abnormal anatomy from medical images. The new algorithm is a form of spatially varying statistical classification, in which an explicit anatomical template is used to moderate the segmentation obtained by statistical classification. The algorithm consists of an iterated sequence of spatially varying classification and nonlinear registration, which forms an adaptive, template moderated (ATM), spatially varying statistical classification (SVC). Classification methods and nonlinear registration methods are often complementary, both in the tasks where they succeed and in the tasks where they fail. By integrating these approaches the new algorithm avoids many of the disadvantages of each approach alone while exploiting the combination. The ATM SVC algorithm was applied to several segmentation problems, involving different image contrast mechanisms and different locations in the body. Segmentation and validation experiments were carried out for problems involving the quantification of normal anatomy (MRI of brains of neonates) and pathology of various types (MRI of patients with multiple sclerosis, MRI of patients with brain tumors, MRI of patients with damaged knee cartilage). In each case, the ATM SVC algorithm provided a better segmentation than statistical classification or elastic matching alone.

Serial intraoperative magnetic resonance imaging of brain shift.

OBJECTIVEA major shortcoming of image-guided navigational systems is the use of preoperatively acquired image data, which does not account for intraoperative changes in brain morphology. The occurrence of these surgically induced volumetric deformations (“brain shift”) has been well established. Maximal measurements for surface and midline shifts have been reported. There has been no detailed analysis, however, of the changes that occur during surgery. The use of intraoperative magnetic resonance imaging provides a unique opportunity to obtain serial image data and characterize the time course of brain deformations during surgery. METHODSThe vertically open intraoperative magnetic resonance imaging system (SignaSP, 0.5 T; GE Medical Systems, Milwaukee, WI) permits access to the surgical field and allows multiple intraoperative image updates without the need to move the patient. We developed volumetric display software (the 3D Slicer) that allows quantitative analysis of the degree and direction of brain shift. For 25 patients, four or more intraoperative volumetric image acquisitions were extensively evaluated. RESULTSSerial acquisitions allow comprehensive sequential descriptions of the direction and magnitude of intraoperative deformations. Brain shift occurs at various surgical stages and in different regions. Surface shift occurs throughout surgery and is mainly attributable to gravity. Subsurface shift occurs during resection and involves collapse of the resection cavity and intraparenchymal changes that are difficult to model. CONCLUSIONBrain shift is a continuous dynamic process that evolves differently in distinct brain regions. Therefore, only serial imaging or continuous data acquisition can provide consistently accurate image guidance. Furthermore, only serial intraoperative magnetic resonance imaging provides an accurate basis for the computational analysis of brain deformations, which might lead to an understanding and eventual simulation of brain shift for intraoperative guidance.

Registration of 3-d intraoperative MR images of the brain using a finite-element biomechanical model

We present a new algorithm for the nonrigid registration of three-dimensional magnetic resonance (MR) intraoperative image sequences showing brain shift. The algorithm tracks key surfaces of objects (cortical surface and the lateral ventricles) in the image sequence using a deformable surface matching algorithm. The volumetric deformation field of the objects is then inferred from the displacements at the boundary surfaces using a linear elastic biomechanical finite-element model. Two experiments on synthetic image sequences are presented, as well as an initial experiment on intraoperative MR images showing brain shift. The results of the registration algorithm show a good correlation of the internal brain structures after deformation, and a good capability of measuring surface as well as subsurface shift. We measured distances between landmarks in the deformed initial image and the corresponding landmarks in the target scan. Cortical surface shifts of up to 10 mm and subsurface shifts of up to 6 mm were recovered with an accuracy of 1 mm or less and 3 mm or less respectively.

Early alteration of structural and functional brain development in premature infants born with intrauterine growth restriction.

Placental insufficiency with fetal intrauterine growth restriction (IUGR) is an important cause of perinatal mortality and morbidity and is subsequently associated with significant neurodevelopmental impairment in cognitive function, attention capacity, and school performance. The underlying biologic cause for this association is unclear. Twenty-eight preterm infants (gestational age 32.5 ± 1.9 wk) were studied by early and term magnetic resonance imaging (MRI). An advanced quantitative volumetric three-dimensional MRI technique was used to measure brain tissue volumes in 14 premature infants with placental insufficiency, defined by abnormal antenatal Doppler measurements and mean birth weights <10th percentile (1246 ± 299 g) (IUGR) and in 14 preterm infants matched for gestational age with normal mean birth weights 1843 ± 246 g (control). Functional outcome was measured at term in all infants by a specialized assessment scale of preterm infant behavior. Premature infants with IUGR had a significant reduction in intracranial volume (mean ± SD: 253.7 ± 29.9 versus 300.5 ± 43.5 mL, p < 0.01) and in cerebral cortical gray matter (mean ± SD: 77.2 ± 16.3 versus 106.8 ± 24.6 mL, p < 0.01) when measured within the first 2 wk of life compared with control premature infants. These findings persisted at term with intracranial volume (mean ± SD: 429.3 ± 47.9 versus 475.9 ± 53.4 mL, p < 0.05) and cerebral cortical gray matter (mean ± SD: 149.3 ± 29.2 versus 189 ± 34.2 mL, p < 0.01). Behavioral assessment at term showed a significantly less mature score in the subsystem of attention-interaction availability in IUGR infants (p ± 0.01). Cerebral cortical gray matter volume at term correlated with attention-interaction capacity measured at term (r = 0.45, p < 0.05). These results suggest that placental insufficiency with IUGR have specific structural and functional consequences on cerebral cortical brain development. These findings may provide insight into the structural-functional correlate for the developmental deficits associated with IUGR.

Late Gestation Cerebellar Growth Is Rapid and Impeded by Premature Birth

Objective. Cognitive impairments and academic failure are commonly reported in survivors of preterm birth. Recent studies suggest an important role for the cerebellum in the development of cognitive and social functions. The objective of this study was to examine the impact of prematurity itself, as well as prematurity-related brain injuries, on early postnatal cerebellar growth with quantitative MRI. Methods. Advanced 3-dimensional volumetric MRI was performed and cerebellar volumes were obtained by manual outlining in preterm (<37 weeks) and healthy term-born infants. Intracranial and total brain volumes were also calculated. Results. A total of 169 preterm and 20 healthy full-term infants were studied; 145 had preterm MRI (pMRI), 75 had term MRI (tMRI), and 51 underwent both pMRI and tMRI. From 28 weeks’ postconceptional age to term, mean cerebellar volume (177%) in preterm infants increased at a much faster rate than did mean intracranial (110%) or mean brain (107%) volumes. Smaller cerebellar volume was significantly related to lower gestational age at birth and to intracranial and total brain volumes. Mean cerebellar volume of preterm infants at tMRI was significantly smaller than the volumes of term-born infants. Cerebellar growth impairment was correlated strongly with associated brain injuries, even in the absence of direct cerebellar injury. Conclusions. Our data suggest that the growth of the immature cerebellum is particularly rapid during late gestation. However, this accelerated growth seems to be impeded by premature birth and associated brain injury. The long-term neurodevelopmental disabilities seen in survivors of premature birth may be attributable in part to impaired cerebellar development.