Quantitative ultrasonography of the periventricular white and grey matter of the developing brain

Effect of Preterm Birth on Echogenicity in Basal Ganglia

In this study, the influence of prematurity on echogenicity of deep gray matter at 30-wk corrected age was assessed using ultrasound measurements. In an observational cohort study, ultrasound scans of 224 extremely preterm infants were prospectively collected. Gray values were assessed in putamen and globus pallidus. Intra- and inter-observer reliability was analyzed and showed excellent agreement. The globus pallidus to putamen ratio was significantly related to gestational age at birth, adjusted regression coefficient in points per wk: 1.28 (95% confidence interval [CI]: 0.38-2.19) for left and 2.12 (95% CI: 1.23-3.02) for right-side images. At 30-wk corrected age this was still the case, adjusted regression coefficient: 0.45 (95% CI: -0.57 to 1.47) for left and 1.29 (95% CI: 0.10-2.48) for right. The putamen is more hyperechoic with lower gestational age. Measuring ultrasound gray values in deep gray matter seems highly reproducible. Prematurity shows a negative correlation with echogenicity of the putamen, this persists at 30-wk corrected age, suggesting altered maturation.

Quantitative analysis of cranial ultrasonographic periventricular echogenicity in relation to early neuromotor development in preterm infants

BACKGROUND

Periventricular white matter (WM) hyperechoic flares that do not evolve into cystic lesion(s) are frequently encountered on cranial ultrasonography (CUS) of preterm infants. Subjective interpretation of its presence, however, is challenging and its association with maturation and neurodevelopment remains undefined.

OBJECTIVES

To determine the relationship between quantitative WM echogenicity and postnatal and postmenstrual ages and the relationship between quantitative WM echogenicity and neuromotor development at term equivalent.

METHODS

We measured the mean pixel brightness intensity at the frontoparietal and parieto-occipital WM, choroid plexus and calvarium bone on sequential neonatal CUS scans of preterm infants born at <34 weeks gestation. The relative echogenicity (RE) was derived by dividing the mean WM echogenicity to that of the choroid plexus (RE(CP)) or bone (RE(BN)). The Lacey Assessment of the Preterm Infant was administered before discharge.

RESULTS

58 preterm infants (the mean gestational age 30.6±2.3 weeks and the mean birth weight 1211.9±224.7 g) were included. The RE(CP) of the frontoparietal WM decreased significantly with advancing postnatal and postmenstrual ages (r=-0.4, p<0.0001). The RE(BN) values of the frontoparietal and parieto-occipital WM during intermediate and late predischarge CUS studies, respectively, were significantly associated with neuromotor status at term (p<0.05). The RE(CP) and RE(BN) measured during the first week of life were not associated with neuromotor status at term.

CONCLUSIONS

Quantitative measurements of the periventricular WM echogenicity are feasible in neonatal CUSs of premature infants and may reflect microstructural developmental changes. An optimal echogenicity quantification technique and its correlation with long-term outcome remain to be determined.

Correlation between a semiautomated method based on ultrasound texture analysis and standard ultrasound diagnosis using white matter damage in preterm neonates as a model

OBJECTIVES

Diagnosis of white matter damage by cranial ultrasound imaging is still subject to interobserver variability and has limited sensitivity for predicting abnormal neurodevelopment later in life. In this study, we evaluated the ability of a semiautomated method based on ultrasound texture analysis to identify patterns that correlate with the ultrasound diagnosis of white matter damage.

METHODS

The study included 44 very preterm neonates born at a median gestational age of 29 weeks 3 days (range, 26 weeks-31 weeks 6 days). Patients underwent cranial ultrasound scans within 1 week of birth and between 14 and 31 days of life. Periventricular leukomalacia was diagnosed by experienced clinicians on the 14- to 31-day scan according to standard criteria. To perform the texture analysis, 4 regions of interest were delineated in stored images: left and right periventricular areas and choroid plexuses. A classification algorithm was developed on the basis of the best combination of texture coefficients to correlate with the clinical diagnosis, and the ability of this algorithm to predict a later diagnosis of periventricular leukomalacia on the first scan was evaluated using a leave-one-out cross-validation.

RESULTS

Periventricular leukomalacia was diagnosed by the standard procedure in 14 of 44 neonates. The texture classification algorithm performed on the first scan could identify cases with a later diagnosis of periventricular leukomalacia with sensitivity of 100% and accuracy of 97.7%.

CONCLUSIONS

These data support the notion that semiautomated quantitative ultrasound analysis achieves early identification of changes in subclinical stages and warrant further investigation of the role of ultrasound texture analysis methods to improve early detection of neonatal brain damage.

Segmentation of white matter flaring areas in ultrasound images of very-low-birth-weight preterm infants

In this article, we present an interactive algorithm segmenting white brain matter, visible as hyperechoic flaring areas in ultrasound (US) images of preterm infants with periventricular leukomalacia (PVL). The algorithm combines both the textural properties of pathological brain tissue and mathematical morphology operations. An initial flaring area estimate is derived from a multifeature multiclassifier tissue texture classifier. This area is refined based on the structural properties of the choroid plexus, a brain feature known to have characteristics similar to flaring. Subsequently, a combination of a morphological closing, gradient and opening by reconstruction operation determines the final flaring area boundaries. Experimental results are compared with a gold standard constructed from manual flaring area delineations of 12 medical experts. In addition, we compared our algorithm to an existing active contour method. The results show our technique agrees to the gold standard with statistical significance and outperforms the existing method in accuracy. Finally, using the flaring area as a criterion we improve the sensitivity of PVL detection up to 98% as compared with the state of the art.

Integrated backscatter of the brain of preterm infants

We measured integrated backscatter (IBS) in the brain of preterm infants using acoustic ultrasound. The study group consisted of 25 preterm infants (gestational age, 32.4+/-2.5 weeks; birth weight, 1488+/-422 g). In parasagittal scans through the posterior horn of the lateral ventricle, regions of interest (ROI) were positioned in the cerebral white matter near the posterior horn (P), anterior horn (A) of the lateral ventricle, and the thalamus (T). IBS of the ROI was measured and IBS of P minus T (P-T) and IBS of A minus T (A-T) were calculated. A-T was greater than P-T. A-T and P-T decreased with increasing gestational age and birth weight. These changes may represent maturation of the cerebrum. A-T or P-T may be useful parameters of cerebral tissue characterization.

Selecting and assessing quantitative early ultrasound texture measures for their association with cerebral palsy

Cerebral palsy (CP) develops as a consequence of white matter damage (WMD) in approximately one out of every 10 very preterm infants. Ultrasound (US) is widely used to screen for a variety of brain injuries in this patient population, but early US often fails to detect WMD. We hypothesized that quantitative texture measures on US images obtained within one week of birth are associated with the subsequent development of CP. In this retrospective study, using images from a variety of US machines, we extracted unique texture measures by means of adaptive processing and high resolution feature enhancement. We did not standardize the images, but used patients as their own controls. We did not remove speckle, as it may contain information. To test our hypothesis, we used the "random forest" algorithm to create a model. The random forest classifier achieved a 72% match to the health outcome of the patients (CP versus no CP), whereas designating all patients as having CP would have resulted in 53% error. This suggests that quantitative early texture measures contain diagnostic information relevant to the development of CP.

Ultrasonic radio-frequency spectrum analysis of normal brain tissue

Acoustic tissue properties can be estimated using texture and/or spectral parameter analysis. Spectral analysis is based on the rf-signals whose frequency-content is commonly neglected in conventional B-mode imaging. Attenuation and backscatter values of normal brain tissue were analyzed. Unprocessed rf-data of 20 patients were sampled intraoperatively after craniotomy using a modified conventional ultrasonic device (Hitachi CS 9600) and analyzed off-line by a custom-made software routine. Before parameter estimation, influences of the diffraction pattern were compensated by means of a correction function obtained using a tissue-mimicking phantom. Attenuation of white matter showed a linear frequency dependence with a slope of 0.94 +/- 0.13 dB cm(-1) MHz(-1). The spectral slope was determined using 10 distinct frequencies between 2.5 and 5.75 MHz. Backscattering properties were analyzed by determining the power spectral density (PSD) and a relative backscatter coefficient (rel BSC) against the values derived from the tissue-mimicking phantom. PSD and rel BSC values were frequency-dependent, with highest PSD values at the probe's center frequency (-75.69 +/- 8.26 dB V(2) Hz(-1)). The corresponding rel BSC value at 5 MHz was determined as 15.39 +/- 8.26 dB. Finally, backscatter coefficients (BSC) of brain tissue were computed using the known BSC of the phantom. The data provided in this study are meant to serve as a base for intended future characterization of brain tissue that potentially allows intraoperative differentiation between normal and pathologic areas and therefore provides the surgeon with additional information for defining the extent of resection in brain more precisely.

A tissue-specific adaptive texture filter for medical ultrasound images

The interpretation of ultrasound images remains a difficult task and the opinion of different doctors is generally not unequivocal. Therefore, there is a growing interest in the field of computer-aided diagnosis. In the field of medical image processing, computer-aided diagnosis includes image enhancement to facilitate visual interpretation, automatic indication of affected areas, organs and other regions of medical interest, the performance of automatic measurements and image registration. In this article, we introduce a new algorithm for ultrasound image enhancement that employs a multivariate texture classifier based on the co-occurrence matrix, which, in combination with an adaptive texture smoothing filter, is used to enhance the visual difference between and improve boundary detection between healthy neonatal brain tissue and tissue affected by periventricular leukomalacia. For a quantitative comparison, we delineate the periventricular leukomalacia-affected regions with two different active contours before and after processing 10 images with the proposed technique and several speckle filters from the literature. The semi-automatic delineations thus obtained are compared with the manual delineations of a neonatologist. In all cases, the average delineation achieved with the proposed technique is closer to that of the manual expert delineation than when the images are processed with the other techniques.

Calibrated parametric medical ultrasound imaging

The goal of this study was to develop a calibrated on-line technique to extract as much diagnostically-relevant information as possible from conventional video-format echograms. The final aim is to improve the diagnostic potentials of medical ultrasound. Video-output images were acquired by a frame grabber board incorporated in a multiprocessor workstation. Calibration images were obtained from a stable tissue-mimicking phantom with known acoustic characteristics. Using these images as reference, depth dependence of the gray level could fairly be corrected for the transducer performance characteristics, for the observer-dependent equipment settings and for attenuation in the examined tissues. Second-order statistical parameters still displayed some nonconsistent depth dependencies. The results obtained with two echoscanners for the same phantom were different; hence, an a posteriori normalization of clinical data with the phantom data is indicated. Prior to processing of clinical echograms,. the anatomical reflections and echoless voids were removed automatically. The final step in the preprocessing concerned the compensation of the overall attenuation in the tissue. A 'sliding window' processing was then applied to a region of interest (ROI) in the 'back-scan converted' images. A number of first and second order statistical texture parameters and acoustical parameters were estimated in each window and assigned to the central pixel. This procedure results in a set of new 'parametric' images of the ROI, which can be inserted in the original echogram (gray value, color) or presented as a color overlay. A clinical example is presented for illustrating the potentials of the developed technique. Depending on the choice of the parameters, four full resolution calibrated parametric images can be calculated and simultaneously displayed within 5 to 20 seconds. In conclusion, an on-line technique has been developed to estimate acoustic and texture parameters with a reduced equipment dependence and to display acoustical and textural information that is present in conventional echograms.