U. Räth

Computerized Ultrasound B-Scan Texture Analysis of Experimental Fatty Liver Disease: Influence of Total Lipid Content and Fat Deposit Distribution

Statistical pattern recognition procedures allow a quantitative description of ultrasound-B-scan image texture. According to well-established animal models, different types of fatty liver disease were induced in female Wistar rats. For the correlation of the computerized ultrasound image with its underlying histology a variable tissue model based on histomorphological data, texture analysis of the histological image and biochemical measurements of total lipid, water and hydroxyprolin content was created. Whereas a regional arrangement of large fat deposits leads to a significant increase in the “mean grey level” (measure of image brightness) of the ultrasound-B-scan image, there is no difference in image brightness between normal liver tissue and liver steatosis for the tissue model with diffuse homogeneous fatty infiltration. It is demonstrated by multiple linear regression analysis that the “mean grey level” of the ultrasound-B-scan image depends not only on total lipid content but even more on the histomorphological fat deposit distribution.

Echographic tissue characterization in diffuse parenchymal liver disease: Correlation of image structure with histology

Seventy livers were examined in vitro using a computerized ultrasound B-mode data acquisition and analysis system. For tissue characterization, statistical parameters from pattern recognition algorithms describing image brightness and image structure were used. Reference classification based on histopathology as well as on chemical/morphometrical analysis led to the diagnostic classes of normal, fatty liver, fibrosis/cirrhosis and fatty fibrosis/cirrhosis. Comparing the two reference methods for ultrasound tissue characterization, re-classification based on chemical/morphometrical analysis resulted in a significant increase in diagnostic accuracy. The strong correlations between statistical ultrasound image parameters and morphometrical features reflect the relevance of our statistical approach to ultrasound tissue characterization.

Ultrasonic Tissue Characterization Using a Diagnostic Expert System

Conventional gray scale B-mode ultrasound is a commonly applied and useful method for a large number of diseases. The operator dependency, however, is limiting its clinical usefulness as well as the reproducability of an examination.

Ultrasonic Tissue Characterization By Texture Analysis: Elimination Of Tissue-Independent Factors

When image texture analysis methods are used for ultrasonic tissue characterization, the discrimination results obtained by statistical pattern discrimination methods must be interpreted carefully in order to avoid pseudo-discriminations due to differences in exa-mination procedures and system settings. This study examines the dependence of popular texture analysis methods on transducer-specific diffraction characteristics, B-mode image reconstruction and sampling factors, i.e. size and position of the selected Region-of-Interest. It is shown that image analysis should always be based on diffraction-corrected ultra-sound signals. In large-organ applications, e.g. liver, polar reconstruction yielded more stable results than cartesian reconstruction, especially when texture measures from the greylevel runlength matrices or power spectrum are used. Analyzing clinical and synthesized ultrasound images, we found that the first-order greylevel statistics: Mean greylevel, skewness and excess as well as the second-order sta-tistics: Correlation of greylevel cooccurrences proved to be stable with respect to tissue-independent factors as well as sufficiently sensitive to tissue differences.

System zum Vergleich von sonographischer Textur und histologischem Aufbau

Mit Hilfe der Sonographie konnen aufgrund der Wechselwirkungen zwischen Gewebe und Ultraschallwellen Organstrukturen und umschriebene Prozesse makroanatomisch bis in Millimeterbereiche abgegrenzt werden. Die Ultraschallbildtextur, d. h. die Anordnung und Helligkeit der Bildechos, wird dagegen durch den feingeweblichen Aufbau bestimmt. In diesem Artikel wird ein computerunterstutztes System vorgestellt, das eine Verbindung zwischen der quantitativen Texturanalyse des Ultraschallbildes und der quantitativen Analyse eines histologischen Gewebeschnittes ermoglicht (Abb. 1).

Diffraktion und ihre Bedeutung für die computerunterstützte B-Bild-Analyse

Die Parameter aus den Grauwerthistogrammen werden als sehr wichtige Parameter fur die Gewebsdifferenzierung diffuser Parenchymveranderungen angesehen [1].

Computer-Assisted Echographic Tissue Characterization in Tumor Diagnostics

The echographic characterization of tissue is open to three different biophysical approaches: 1. The analysis of the radio-frequency signal (RF-signal) 2. The analysis of the video-A signal 3. The analysis of the B-scan A hardware-software system for the acquisition and evaluation of ultrasonic data was developed in our institute. This system permitted the quantitative analysis of a series of A-scans by means of distinct statistical parameter sets leading to the differentiation of normal tissue from cirrhotic and metastatic liver [1,3].

Geräte und Untersuchungsgang

Hinsichtlich Abtastvorgang, Geschwindigkeit des Bildaufbaus, Bildfolgefre-quenz, und Scanart kann man die Ultraschall-B-Scan-Gerate in der in Abb. 1.1 angegebenen Weise unterscheiden. Zusatzlich gibt es dann noch die Moglich-keit des direkten Kontaktscans, wobei das Gerat auf die Hautoberflache aufge-setzt wird, und die Moglichkeit, eine Wasservorlaufstrecke zwisehen den Schallkopf und die Korperoberflache zu schalten.

An Image Analysis Workstation for Ultrasonic Tissue Characterization

Conventional ultrasound imaging is a frequently applied and useful method for the diagnosis of a wide variety of diseases. However, in the literature there is some discussion about considerable operator variation being observed originating from operator dependency and leading to varying diagnostic accuracies. Therefore, to overcome the reproducability problem, as well as to inhance the diagnostic value of the imaging technique, it is clinically useful to explore the computer assisted approach to tissue characterization.