Search

Anti-angiogenic therapy and immune checkpoint inhibition are novel treatment strategies for patients with renal cell carcinoma. Various components and structures of the tumor microenvironment are potential predictive biomarkers and also attractive treatment targets. Macrophages, tumor infiltrating lymphocytes, vascular and lymphatic vessels represent an important part of the tumor immune environment, but their functional phenotypes and relevance for clinical outcome are yet ill defined. We applied Tissue Phenomics methods including image analysis for the standardized quantification of specific components and structures within the tumor microenvironment to profile tissue sections from 56 clear cell renal cell carcinoma patients. A characteristic composition and unique spatial relationship of CD68+ macrophages and tumor infiltrating lymphocytes correlated with overall survival. An inverse relationship was found between vascular (CD34) and lymphatic vessel (LYVE1) density. In addition, outcome was significantly better in patients with high blood vessel density in the tumors, whereas increased lymphatic vessel density in the tumors was associated with worse outcome. The Tissue Phenomics imaging analysis approach allowed visualization and simultaneous quantification of immune environment components, adding novel contextual information, and biological insights with potential applications in treatment response prediction.

RATIONALE AND OBJECTIVES To evaluate a new approach to establish compliance of segmentation tools with the computed tomography volumetry profile of the Quantitative Imaging Biomarker Alliance (QIBA); and determine the statistical exchangeability between real and simulated lesions through an international challenge. MATERIALS AND METHODS The study used an anthropomorphic phantom with 16 embedded physical lesions and 30 patient cases from the Reference Image Database to Evaluate Therapy Response with pathologically confirmed malignancies. Hybrid datasets were generated by virtually inserting simulated lesions corresponding to physical lesions into the phantom datasets using one projection-domain-based method (Method 1), two image-domain insertion methods (Methods 2 and 3), and simulated lesions corresponding to real lesions into the Reference Image Database to Evaluate Therapy Response dataset (using Method 2). The volumes of the real and simulated lesions were compared based on bias (measured mean volume differences between physical and virtually inserted lesions in phantoms as quantified by segmentation algorithms), repeatability, reproducibility, equivalence (phantom phase), and overall QIBA compliance (phantom and clinical phase). RESULTS For phantom phase, three of eight groups were fully QIBA compliant, and one was marginally compliant. For compliant groups, the estimated biases were -1.8 ± 1.4%, -2.5 ± 1.1%, -3 ± 1%, -1.8 ± 1.5% (±95% confidence interval). No virtual insertion method showed statistical equivalence to physical insertion in bias equivalence testing using Schuirmann's two one-sided test (±5% equivalence margin). Differences in repeatability and reproducibility across physical and simulated lesions were largely comparable (0.1%-16% and 7%-18% differences, respectively). For clinical phase, 7 of 16 groups were QIBA compliant. CONCLUSION Hybrid datasets yielded conclusions similar to real computed tomography datasets where phantom QIBA compliant was also compliant for hybrid datasets. Some groups deemed compliant for simulated methods, not for physical lesion measurements. The magnitude of this difference was small (

Gastric cancer with lymphoid stroma (GCLS) is characterized by prominent stromal infiltration of T-lymphocytes. The aim of this study was to investigate GCLS biology through analysis of clinicopathological features, EBV infection, microsatellite instability (MSI), immune gene-expression profiling and PD-L1 status in neoplastic cells and tumor immune microenvironment. Twenty-four GCLSs were analyzed by RNA in situ hybridization for EBV (EBER), PCR/fragment analysis for MSI, immunohistochemistry (PD-L1, cytokeratin, CD3, CD8), co-immunofluorescence (CK/PD-L1, CD68/PD-L1), NanoString gene-expression assay for immune-related genes and PD-L1 copy number alterations. CD3+ and CD8+ T-cell densities were calculated by digital analysis. Fifty-four non-GCLSs were used as control group. GCLSs displayed distinctive clinicopathological features, such as lower pTNM stage (p = 0.02) and better overall survival (p = 0.01). EBV+ or MSI-high phenotype was found in 66.7 and 16.7% cases, respectively. GCLSs harbored a cytotoxic T-cell-inflamed profile, particularly at the invasive front of tumors (p

Tissue Phenomics is the discipline of mining tissue images to identify patterns that are related to clinical outcome providing potential prognostic and predictive value. This involves the discovery process from assay development, image analysis, and data mining to the final interpretation and validation of the findings. Importantly, this process is not linear but allows backward steps and optimization loops over multiple sub-processes. We provide a detailed description of the Tissue Phenomics methodology while exemplifying each step on the application of prostate cancer recurrence prediction. In particular, we automatically identified tissue-based biomarkers having significant prognostic value for low- and intermediate-risk prostate cancer patients (Gleason scores 6–7b) after radical prostatectomy. We found that promising phenes were related to CD8(+) and CD68(+) cells in the microenvironment of cancerous glands in combination with the local micro-vascularization. Recurrence prediction based on the selected phenes yielded accuracies up to 83% thereby clearly outperforming prediction based on the Gleason score. Moreover, we compared different machine learning algorithms to combine the most relevant phenes resulting in increased accuracies of 88% for tumor progression prediction. These findings will be of potential use for future prognostic tests for prostate cancer patients and provide a proof-of-principle of the Tissue Phenomics approach.

Classifying whole-slide images of prostate cancer resections to derive an accurate prognosis for tumor progression is a highly challenging problem. We here introduce a novel type of high-level features which operate on the gland level for the application in computer-aided prognosis using a Tissue Phenomics approach. Since tissue architecture, and in particular, the gland distribution possibly provide information on the aggressiveness of the individual tumor, our features exploit the spatial relationship of different gland types. Glands are classified into cancerous and healthy glands, and also, into morphological classes based on size and shape. Co-occurrences of the classified glands are quantified and Haralick-like features are computed based on the derived gland co-occurrence matrices. The resulting gland co-occurrence features are mined to automatically determine the best parametrization. In experiments on whole-slide images it turned out that our novel features allow accurate stratification of patients into the prognostic groups tumor progression and nonprogression outperforming clinical features. Our results indicate a strong correlation of tumor progression with invasion phenotypes.

// Elisa Neubauer 1 , Ralph M. Wirtz 2 , Daniel Kaemmerer 3 , Maria Athelogou 4 , Lydia Schmidt 1 , Jorg Sanger 5 , Amelie Lupp 1 1 Institute of Pharmacology and Toxicology, Jena University Hospital, Friedrich Schiller University, Jena, Germany 2 STRATIFYER Molecular Pathology GmbH Koln, Koln, Germany 3 Department of General and Visceral Surgery, Zentralklinik Bad Berka, Bad Berka, Germany 4 DEFINIENS AG, Munchen, Germany 5 Laboratory of Cytology and Pathology, Bad Berka, Germany Correspondence to: Elisa Neubauer, email: elisa.specht@med.uni-jena.de Keywords: Ki-67, topoisomerase 2 alpha, RacGAP1, lung neuroendocrine neoplasms, IHC/RT-qPCR Received: March 13, 2016 Accepted: May 16, 2016 Published: May 31, 2016 ABSTRACT The classification of bronchopulmonary neuroendocrine neoplasms (BP-NEN) into four tumor entities (typical carcinoids (TC), atypical carcinoids (AC), small cell lung cancers (SCLC), large cell neuroendocrine lung carcinomas (LCNEC)) is difficult to perform accurately, but important for prognostic statements and therapeutic management decisions. In this regard, we compared the expression of three proliferation markers, Ki-67, Topoisomerase II alpha (TOP2A), and RacGAP1, in a series of tumor samples from 104 BP-NEN patients (24 TC, 21 AC, 52 SCLC, 7 LCNEC) using different evaluation methods (immunohistochemistry (IHC): Average evaluation, Hotspot evaluation, digital image analysis; RT-qPCR). The results indicated that all three markers had increased protein and mRNA expression with poorer differentiation and correlated well with each other, as well as with grading, staging, and poor survival. Compared with Ki-67 and TOP2A, RacGAP1 allowed for a clearer prognostic statement. The cut-off limits obtained for Ki-67-Average (IHC) were TC-AC 1.5, AC-SCLC 19, and AC-LCNEC 23.5. The Hotspot evaluation generated equal to higher, the digital image analysis generally lower between-entity cut-off limits. All three markers enabled a clear-cut differentiation between the BP-NEN entities, and all methods evaluated were suitable for marker assessment. However, to define optimal cut-off limits, the Ki-67 evaluation methods should be standardized. RacGAP1 appeared to be a new marker with great potential.

Rationale and Objectives Quantifying changes in lung tumor volume is important for diagnosis, therapy planning, and evaluation of response to therapy. The aim of this study was to assess the performance of multiple algorithms on a reference data set. The study was organized by the Quantitative Imaging Biomarker Alliance (QIBA). Materials and Methods The study was organized as a public challenge. Computed tomography scans of synthetic lung tumors in an anthropomorphic phantom were acquired by the Food and Drug Administration. Tumors varied in size, shape, and radiodensity. Participants applied their own semi-automated volume estimation algorithms that either did not allow or allowed post-segmentation correction (type 1 or 2, respectively). Statistical analysis of accuracy (percent bias) and precision (repeatability and reproducibility) was conducted across algorithms, as well as across nodule characteristics, slice thickness, and algorithm type. Results Eighty-four percent of volume measurements of QIBA-compliant tumors were within 15% of the true volume, ranging from 66% to 93% across algorithms, compared to 61% of volume measurements for all tumors (ranging from 37% to 84%). Algorithm type did not affect bias substantially; however, it was an important factor in measurement precision. Algorithm precision was notably better as tumor size increased, worse for irregularly shaped tumors, and on the average better for type 1 algorithms. Over all nodules meeting the QIBA Profile, precision, as measured by the repeatability coefficient, was 9.0% compared to 18.4% overall. Conclusion The results achieved in this study, using a heterogeneous set of measurement algorithms, support QIBA quantitative performance claims in terms of volume measurement repeatability for nodules meeting the QIBA Profile criteria.

Introduction: The composition of different immune cell populations in the tumor microenvironment plays an important role for tumor progression in various cancer types. In particular, tumor-associated macrophages (TAMs) and tumor-infiltrating T cells (TILs) are relevant players. On the other hand, structural changes such as neoangiogenesis triggered by cancerous tumors to increase their nutrient supplies have also been associated with tumor progression. In this work we systematically quantified these factors in prostate cancer (PCa) and correlated them with clinical outcome data using a Tissue Phenomics approach. By investigating the prognostic relevance of TAMs (CD68/CD163), TILs (CD3/CD8), and microvessels (CD34) in the tumor, tumor microenvironment (TME), and stroma we identified strong prognostic markers for PCa recurrence prediction in patients after radical prostatectomy. Methods: In this study, we analyzed a cohort of 90 PCa patients, of whom 40 suffered from tumor progression measured by prostate cancer antigen (PSA) recurrence after prostatectomy. The cohort comprised low- and intermediate-risk PCa patients (Gleason-Score≤7b) since providing a reliable prognosis is particularly difficult for such grades. Tissue sections were immunohistochemically stained using the duplex stains CD68/CD163 for TAMs, CD3/CD8 for TILs, and CK18/p63 to identify and characterize glands as cancerous vs. noncancerous based on their expression level of p63 (in cancerous glands p63 is not expressed). To quantify tumor neoangiogenesis microvessels were stained by CD34. All sections were geometrically aligned per case (virtual multiplexing) to enable coanalysis of stains, and quantified within relevant regions-of-interest (tumor, TME, stroma) using fully automated computational methods (1, 2). In particular, we determined region-specific densities and average distances of TAMs, TILs, and microvessels, as well as ratios of all measures. We systematically analyzed the prognostic power of each measure by optimizing a cutoff with respect to the disease-free survival statistic (log-rank test) using cross-validation to avoid for overfitting. Results: The top-ranking prognostic markers regarding robustness and prediction performance were related to microvessel density combined with immune cell densities. In particular, we found that within the TME, a coverage of CD8(+) cytotoxic T cells larger than 10% of the coverage of CD34(+) microvessels is correlated with a good prognosis and long-term disease-free survival (cross-validated p Conclusion: Our results indicate a considerable prognostic potential of markers combining microvessel density with measures of TAMs and TILs to predict PSA recurrence in PCa. This application shows that systematic analysis as performed by Tissue Phenomics enables discovery of non-obvious combined prognostic markers characterizing the tumor landscape with high potential to improve patient treatment. In future work we aim to validate our findings on additional data from other clinical sites. References: 1. Yigitsoy M, et al. Hierarchical patch-based co-registration of differently stained histopathology slides. Proc SPIE 2017. doi:10.1117/12.2254266. 2. Brieu N, et al. Slide specific models for segmentation of differently stained digital histopathology whole slide images. Proc SPIE 2016. doi:10.1117/12.2208620. Citation Format: Nathalie Harder, Maria Athelogou, Harald Hessel, Alexander Buchner, Christian Stief, Thomas Kirchner, Günter Schmidt, Ralf Huss, Tze Heng Tan. Combination of immune status and tumor microvascularization provides strong prognostic markers for prostate cancer recurrence prediction [abstract]. In: Proceedings of the AACR Special Conference: Prostate Cancer: Advances in Basic, Translational, and Clinical Research; 2017 Dec 2-5; Orlando, Florida. Philadelphia (PA): AACR; Cancer Res 2018;78(16 Suppl):Abstract nr A069.

We present a new approach for computer-aided detection and diagnosis in mammography based on Cognition Network Technology (CNT). Originally designed for image processing, CNT has been extended to also perform context- and knowledge-driven analysis of tabular data. For the first time using this technology, an application was created and evaluated for fully automatic searching of patient cases from a reference database of verified findings. The application aims to support radiologists in providing cases of similarity and relevance to a given query case. It adopts an extensible and knowledge-driven concept as a similarity measure. As a preprocessing step, all input images from more than 400 patients were fully automatically segmented and the resulting objects classified—this includes the complete breast shape, the position of the mammilla, the pectoral muscle, and various potential candidate objects for suspicious mass lesions. For the similarity search, collections of object properties and metadata from many patients were combined into a single table analysis project. Extended CNT allows for a convenient implementation of knowledge-based structures, for example, by meaningfully linking detected objects in different breast views that might represent identical lesions. Objects from alternative segmentation methods are also be considered, so as to collectively become a sufficient set of base-objects for identifying suspicious mass lesions. For 80% of 112 patient cases with suspicious lesions, the system correctly identified at least one corresponding mass lesion as an object of interest. In this database, consisting of 1,024 images from a total of 303 patients, an average of 0.66 false-positive objects per image were detected. An additional testing database contained 480 images from 120 patients, 15 of whom were annotated with suspicious mass lesions. Here, 47% (7 out of 15) of these were detected automatically with 1.13 false-positive objects per image. A diagnosis is predicted for each patient case by applying a majority vote from the reference findings of the ten most similar cases. Two separate evaluation scenarios suggest a fraction of correct predictions of respectively 79 and 76%. Cognition Network Technology was extended to process table data, making it possible to access and relate records from different images and non-image sources, such as demographic patient data or parameters from clinical examinations. A prototypal application enables efficient searching of a patient and image database for similar patient cases. Using concepts of knowledge-driven configuration and flexible extension, the application illustrates a path to a new generation of future CAD systems.

Rationale and objectives Tumor volume change has potential as a biomarker for diagnosis, therapy planning, and treatment response. Precision was evaluated and compared among semiautomated lung tumor volume measurement algorithms from clinical thoracic computed tomography data sets. The results inform approaches and testing requirements for establishing conformance with the Quantitative Imaging Biomarker Alliance (QIBA) Computed Tomography Volumetry Profile. Materials and Methods Industry and academic groups participated in a challenge study. Intra-algorithm repeatability and inter-algorithm reproducibility were estimated. Relative magnitudes of various sources of variability were estimated using a linear mixed effects model. Segmentation boundaries were compared to provide a basis on which to optimize algorithm performance for developers. Results Intra-algorithm repeatability ranged from 13% (best performing) to 100% (least performing), with most algorithms demonstrating improved repeatability as the tumor size increased. Inter-algorithm reproducibility was determined in three partitions and was found to be 58% for the four best performing groups, 70% for the set of groups meeting repeatability requirements, and 84% when all groups but the least performer were included. The best performing partition performed markedly better on tumors with equivalent diameters greater than 40 mm. Larger tumors benefitted by human editing but smaller tumors did not. One-fifth to one-half of the total variability came from sources independent of the algorithms. Segmentation boundaries differed substantially, not ony in overall volume but also in detail. Conclusions Nine of the 12 participating algorithms pass precision requirements similar to what is indicated in the QIBA Profile, with the caveat that the present study was not designed to explicitly evaluate algorithm profile conformance. Change in tumor volume can be measured with confidence to within ±14% using any of these nine algorithms on tumor sizes greater than 10 mm. No partition of the algorithms was able to meet the QIBA requirements for interchangeability down to 10 mm, although the partition comprising best performing algorithms did meet this requirement for a tumor size of greater than approximately 40 mm.

Purpose: Breast cancer is composed of phenotypically diverse populations of cancer cells. The ability to form breast tumors has been shown by in vitro/in vivo studies to be restricted to epithelial tumor cells with CD44+/CD24−/low characteristics. Validation of these findings with respect to detection in clinical samples, prognosis, and clinical relevance is in demand. Experimental Design: We investigated breast cancer tissues for the prevalence of CD44+/CD24−/low tumor cells and their prognostic value. The study included paraffin-embedded tissues of 136 patients with and without recurrences. In addition, a breast cancer progression array with normal, carcinoma in situ, and carcinoma tissues was analyzed. We applied double-staining immunohistochemistry for the detection of CD44+/CD24−/low cells. Evaluation was by microscopic pathologic inspection and automated image analysis. Results: CD44+/CD24−/low cells ranged from 0% to 40% in normal breast and from 0% to 80% in breast tumor tissues. The prevalence of CD44+/CD24−/low tumor cells in 122 tumors was ≤10% in the majority (78%) of cases and >10% in the remainder. There was no significant correlation between CD44+/CD24−/low tumor cell prevalence and tumor progression. Although recurrences of tumors with high percentages of CD44+/CD24−/low tumor cells were mainly distant, preferably osseous metastasis, there was no correlation with the event-free and overall survival. There was no influence on the response to different treatment modalities. Conclusions: Our findings suggest that the prevalence of CD44+/CD24−/low tumor cells in breast cancer may not be associated with clinical outcome and survival but may favor distant metastasis.

Background: Manual evaluation of somatostatin receptor (SSTR) immunohistochemistry (IHC) is a time-consuming and cost-intensive procedure. Aim of the study was to compare manual evaluation of SSTR subtype IHC to an automated software-based analysis, and to in-vivo imaging by SSTR-based PET/CT. Methods: We examined 25 gastroenteropancreatic neuroendocrine tumor (GEP-NET) patients and correlated their in-vivo SSTR-PET/CT data (determined by the standardized uptake values SUVmax,-mean) with the corresponding ex-vivo IHC data of SSTR subtype (1, 2A, 4, 5) expression. Exactly the same lesions were imaged by PET/CT, resected and analyzed by IHC in each patient. After manual evaluation, the IHC slides were digitized and automatically evaluated for SSTR expression by Definiens XD software. A virtual IHC score “BB1” was created for comparing the manual and automated analysis of SSTR expression. Results: BB1 showed a significant correlation with the corresponding conventionally determined Her2/neu score of the SSTR-subtypes 2A (rs: 0.57), 4 (rs: 0.44) and 5 (rs: 0.43). BB1 of SSTR2A also significantly correlated with the SUVmax (rs: 0.41) and the SUVmean (rs: 0.50). Likewise, a significant correlation was seen between the conventionally evaluated SSTR2A status and the SUVmax (rs: 0.42) and SUVmean (rs: 0.62).Conclusion: Our data demonstrate that the evaluation of the SSTR status by automated analysis (BB1 score), using digitized histopathology slides (“virtual microscopy”), corresponds well with the SSTR2A, 4 and 5 expression as determined by conventional manual histopathology. The BB1 score also exhibited a significant association to the SSTR-PET/CT data in accordance with the high affinity profile of the SSTR analogues used for imaging.

Because of the very special type of contrast in phase-contrast images, it is almost impossible to perform fully automated single-cell analysis and quantification successfully. Because fluorescent dyes are highly toxic, phase-contrast images are commonly used to monitor live cells. In this paper, we present a method for the fully automated segmentation, classification and quantification of individual cell morphology in phase-contrast images. We calculate the confluence of the cell population and quantify the degree of toxic damage to each individual cell following phenol incubation. The results are then compared to standard cytotoxicity assays.

This paper presents a method for the fully automated segmentation, classification and quantification of the morphology of individual cells in phase-contrast images. We utilize this method to calculate automatically the confluence of the corresponding cell culture and to quantify the degree of toxic damage of each individual cell after phenol incubation. These results are compared to standard cytotoxicity assays.

Heutzutage sind bildgebende Verfahren aus medizinischen Untersuchungen nicht mehr wegzudenken. Diverse Methoden – basierend auf dem Einsatz von Ultraschallwellen, Rontgenstrahlung, Magnetfeldern oder Lichtstrahlen – werden dabei spezifisch eingesetzt und liefern umfangreiches Datenmaterial uber den Korper und sein Inneres. Anhand von Mikroskopieaufnahmen aus Biopsien konnen daruber hinaus Daten uber die morphologische Eigenschaften von Korpergeweben gewonnen werden. Aus der Analyse all dieser unterschiedlichen Arten von Informationen und unter Konsultation weiterer klinischer Untersuchungen aus diversen medizinischen Disziplinen kann unter Berucksichtigung von Anamnesedaten ein „Gesamtbild“ des Gesundheitszustands eines Patienten erstellt werden. Durch die Flut der erzeugten Bilddaten kommt der Bildverarbeitung im Allgemeinen und der Bildanalyse im Besonderen eine immer wichtigere Rolle zu. Gerade im Bereich der Diagnoseunterstutzung, der Therapieplanung und der bildgefuhrten Chirurgie bilden sie Schlusseltechnologien, die den Forschritt nicht nur auf diesen Gebieten masgeblich vorantreiben.

Automated digital microscopy is established in biology, pathology, and in drug discovery and development as a means to screen cell assays and tissue slides in high content and high throughput workflows for target identification and validation. It is further employed for analysis of tissue slides from biopsies. Simulations are also used in medical and systems biology to produce alphanumeric data that represents the structure, morphology, behavior and interactions of biological objects such as cells, cell organelles and proteins. Multidimensional digital image data is generated, such as multilayer confocal imagery or alphanumeric simulation results, so as to span both domains, with simulation results forming an input for experimentation and vice versa. In order to understand the structure, function and dynamics of proteins, cell organelles, cells, tissue, organs etc. detailed morphological quantification of objects within these images is needed. We present application results of an object- and context-based image analysis method, which employs domain-knowledge and context information to analyze multidimensional image data automatically.

Heutzutage sind bildgebende Verfahren aus medizinischen Untersuchungen nicht mehr wegzudenken. Diverse Methoden — basierend auf dem Einsatz von Ultraschallwellen, Rontgenstrahlung, Magnetfeldern oder Lichtstrahlen — werden dabei spezifisch eingesetzt und liefern umfangreiches Datenmaterial uber den Korper und sein Inneres. Anhand von Mikroskopieaufnahmen aus Biopsien konnen daruber hinaus Daten uber die morphologische Eigenschaften von Korpergeweben gewonnen werden. Aus der Analyse all dieser unterschiedlichen Arten von Informationen und unter Konsultation weiterer klinischer Untersuchungen aus diversen medizinischen Disziplinen kann unter Berucksichtigung von Anamnesedaten ein “Gesamtbild„ des Gesundheitszustands eines Patienten erstellt werden. Durch die Flut der erzeugten Bilddaten kommt der Bildverarbeitung im Allgemeinen und der Bildanalyse im Besonderen eine immer wichtigere Rolle zu. Gerade im Bereich der Diagnoseunterstutzung, der Therapieplanung und der bildgefuhrten Chirurgie bilden sie Schlusseltechnologien, die den Forschritt nicht nur auf diesen Gebieten masgeblich vorantreiben.

In dieser Arbeit stellen wir einen voll-automatisierten Algorithmus vor, der in der Lage ist, die Struktur des Corpus Callosum aus sagittalen Schichten von T1-gewichteten kernspintomographischen Datensatzen des menschlichen Gehirns zu segmentieren. Die Segmentierungsergebnisse werden dabei fur die Untersuchung morphometrischer Merkmale in einer Studie zur Schizophrenie in definierte Abschnitte unterteilt, um sie im weiteren Verlauf statistisch auswerten zu konnen. Der Algorithmus wurde unter Zuhilfenahme der Cognition Network Technologie implementiert, die eine regelbasierte und kontextsensitive Handhabung der Bilddaten erlaubt, und dabei nur wenige Voraussetzungen uber die Beschaffenheit und Qualitat der zu verarbeitenden Datens atze macht. Das Verfahren scheitert im Rahmen einer Testreihe bei einem von 50 Datensatzen und erzielt ansonsten einen Dice-Koeffizienten von 0,97 im Vergleich zu manuell segmentierten Ergebnissen.

Digital mammography and comprehensive breast cancer screening approaches have led to the generation of a vast amount of image data. Since the visual inspection of a large set of images is expensive and to some extend also subjective, new methods for fully automated mammography image analysis are needed. The Definiens Cognition Network Technology (CNT) solves the image analysis problem by simulating human cognition processes using knowledge based and context dependent processing. It represents processed image data, image processing methods, and image objects and their definitions in a unified model which incorporates elements from semantic networks, description logics and functional programming. We present first steps towards a successful application of this technology on automated detection of masses and calcifications according to the ACR BI-RADS™ standard.

Die Computerassistierte Diagnose (CAD) in der Mammographie hat durch den europaweiten Aufbau von nationalen Screeningprogrammen in den letzten Jahren stark an Bedeutung gewonnen. Fur die Evaluierung und vor allem fur den Vergleich von CAD Algorithmen sind offentlich zugangliche Referenzdaten notig. Am Fraunhofer IIS wird derzeit eine Referenzdatenbank fur die Mammographie aufgebaut, die die veralteten bestehenden Datenbanken ersetzen bzw. erganzen soll. Dabei wurde auf dem Stand der Geratetechnik entsprechende Bildqualitat und zeitgemase ikonische und textuelle Annotation geachtet. Die Veroffentlichung dieser Referenzdaten wird es Forschungsgruppen erlauben, CAD Algorithmen fur die Mammographie auf umfangreichen und aktuellen Referenzdaten zu evaluieren und ihre Leistungsfahigkeit miteinander zu vergleichen.

Data analysis in general and image analysis in particular require multi-scale approaches when dealing with complex structures. Relational information between structures on different scales needs to be taken into account. In many application fields, automated image interpretation still is a significant bottleneck due to the lack of appropriate image analysis technology. A new approach, Cognition Network Technology, is presented that was developed to handle and analyze complex data. This contribution focuses on how it handles and analyzes image data based on an object oriented, hierarchical and networked data model. A specific programming language allows building a semantic knowledge base that is used to interpreting image data by creating and processing instances of this data model. In many operational analysis tasks the approach has proven to produce reliable results fully automatically. It especially extracts structures of interest even in challenging cases such as low signal to noise ratio images, heterogeneous or variable structures of interest or tasks which include a complex semantic.

Volumetrische Mase des menschlichen Ventrikelsystems sind ein wichtiger Bestandteil zur Beurteilung des Krankheitsverlaufs bei neurodegenerativen Krankheiten, wie z.B der Alzheimerschen Demenz (AD). In dieser Arbeit wird ein Segmentierungs-Algorithmus vorgestellt, der die Lage und das Volumen der Seitenventrikel des menschlichen Gehirns automatisch bestimmen kann, so das Zeit- und Arbeitsaufwand manueller Segmentierungen stark reduziert werden. Die verwendeten Algorithmen verfolgen dabei einen regelbasierten und objektorientieren Ansatz, der hier erstmals fur die Verwendung von 3D-Daten angepast und evaluiert wurde. Die zugrundeliegenden 20 Datensatze entstammen einer AD-Studie, wobei 10 Datensatze von 5 Experten zum Vergleich manuell segmentiert wurden. Die erzielten Ergebnisse liegen im Bereich der Inter-Rater-Variabilitat, und der voll-automatische Algorithmus scheitert nur in 1/20 der Falle.

Licht- und elektronenmikroskopischen Bilddaten sind heute etablierte Informationsmedien in der Medizin und Biologie. Multispektrale digitale Bilder mit komplexen Inhalten werden in grosen Mengen halb- und vollautomatisch erzeugt. Eine Analyse im Sinne einer Automatisierung der Extraktion von vernetztem Wissen verlangt nach neuen Methoden. Die Spezialisierung pixelbasierte Losungen ermoglicht oft automatische Bildanalysen, bleibt aber nichts desto trotz in gewissen Rahmen auf spezifische Applikationen beschrankt. Der fraktal hierarchische prozes- und objektbasierte Ansatz der Cognition Networks, ist in der Lage komplexe Systeme zu analysieren und zu simulieren. Dabei wird ein neues selbstahnliches dynamisches Objektmodell benutzt.