Thomas J. Olson

Real-time vergence control for binocular robots

In binocular systems,vergence is the process of adjusting the angle between the eyes (or cameras) so that both eyes are directed at the same world point. Its utility is most obvious for foveate systems such as the human visual system, but it is a useful strategy for nonfoveate binocular robots as well. Here, we discuss the vergence problem and outline a general approach to vergence control, consisting of a control loop driven by an algorithm that estimates the vergence error. As a case study, this approach is used to verge the eyes of the Rochester Robot in real time. Vergence error is estimated with the cepstral disparity filter. The cepstral filter is analyzed, and it is shown in this application to be equivalent to correlation with an adaptive prefilter; carrying this idea to its logical conclusion converts the cepstral filter into phase correlation. The demonstration system uses a PD controller in cascade with the error estimator. An efficient real-time implementation of the error estimator is discussed, and empirical measures of the performance of both the disparity estimator and the overall system are presented.

Comparing Depth From Motion With Depth From Binocular Disparity

The accuracy of depth judgments that are based on binocular disparity or structure from motion (motion parallax and object rotation) was studied in 3 experiments. In Experiment 1, depth judgments were recorded for computer simulations of cones specified by binocular disparity, motion parallax, or stereokinesis. In Experiment 2, judgments were recorded for real cones in a structured environment, with depth information from binocular disparity, motion parallax, or object rotation about the y-axis. In both of these experiments, judgments from binocular disparity information were quite accurate, but judgments on the basis of geometrically equivalent or more robust motion information reflected poor recovery of quantitative depth information. A 3rd experiment demonstrated stereoscopic depth constancy for distances of 1 to 3 m using real objects in a well-illuminated, structured viewing environment in which monocular depth cues (e.g., shading) were minimized.

The analysis of clonal diversity and therapy responses using STAT3 mutations as a molecular marker in large granular lymphocytic leukemia

T-cell large granular lymphocytic leukemia and chronic lymphoproliferative disorder of natural killer cells are intriguing entities between benign and malignant lymphoproliferation. The molecular pathogenesis has partly been uncovered by the recent discovery of somatic activating STAT3 and STAT5b mutations. Here we show that 43% (75/174) of patients with T-cell large granular lymphocytic leukemia and 18% (7/39) with chronic lymphoproliferative disorder of natural killer cells harbor STAT3 mutations when analyzed by quantitative deep amplicon sequencing. Surprisingly, 17% of the STAT3-mutated patients carried multiple STAT3 mutations, which were located in different lymphocyte clones. The size of the mutated clone correlated well with the degree of clonal expansion of the T-cell repertoire analyzed by T-cell receptor beta chain deep sequencing. The analysis of sequential samples suggested that current immunosuppressive therapy is not able to reduce the level of the mutated clone in most cases, thus warranting the search for novel targeted therapies. Our findings imply that the clonal landscape of large granular lymphocytic leukemia is more complex than considered before, and a substantial number of patients have multiple lymphocyte subclones harboring different STAT3 mutations, thus mimicking the situation in acute leukemia.

Stereopsis for verging systems

The implications of vergence control and active vision for stereopsis in robots and humans are investigated. It is argued that the geometry of verging systems places strong constraints on the ecological role of stereopsis. In particular, stereopsis is poorly suited to building whole-field depth maps but well suited to gathering information about specific targets of interest in the scene in a coordinate frame referenced to the fixation point. A simple, fast stereo system targeted at the latter goal is presented.<<ETX>>

Fixation-based filtering

Fixation and visual attention are central themes in active vision research, and are closely related. In this paper we discuss one of several ways in which they interact. We describe filtering methods that allow an agent to selectively extract features of the object it is fixating and suppress features of foreground and background objects. The methods are essentially depth filters; they use disparity or motion information to suppress image features that are far from the fixation point in depth. They share a simple computational structure based on the Laplacian pyramid, and are readily amenable to hardware implementation. We present the filters and the properties of fixation geometry that allow them to work, and discuss their behavior. We present methods of implementing them in real time and describe ways of extending them to other features besides depth.

Programming a pipelined image processor

This paper describes a software system that we have developed to simplify the task of programming the DataCube MV20 image processor. The system presents an abstract view of the hardwares capabilities, allowing the programmer to focus on the computation to be performed rather than the manipulations needed to map the computation onto the hardware. Because it is based on an abstract model of the hardware, the system could be supported on other architectures as well. The core of the programming system is VEIL. a C++ library that provides a dataflow abstraction for programming the underlying machine. VEIL represents computations as directed graphs whose nodes are standard image processing operators (add, subtract, convolve, etc), and whose arcs represent communications channels. User programs construct VEIL graphs procedurally, by calling functions that instantiate nodes and link their inputs and outputs as needed. VEIL automatically maps the nodes and arcs of the graph onto the underlying hardware, breaking the graph into subgraphs if the computation is too complex to be performed in a single cycle.

Somatic STAT3 mutations in the Felty syndrome: an implication for a common pathogenesis with large granular lymphocyte leukemia

Felty syndrome is a rare disease defined by neutropenia, splenomegaly, and rheumatoid arthritis. Sometimes the differential diagnosis between Felty syndrome and large granular lymphocyte leukemia is problematic. Recently, somatic STAT3 and STAT5B mutations were discovered in 30–40% of patients with large granular lymphocyte leukemia. Herein, we aimed to study whether these mutations can also be detected in Felty syndrome, which would imply the existence of a common pathogenic mechanism between these two disease entities. We collected samples and clinical information from 14 Felty syndrome patients who were monitored at the rheumatology outpatient clinic for Felty syndrome. Somatic STAT3 mutations were discovered in 43% (6/14) of Felty syndrome patients with deep amplicon sequencing targeting all STAT3 exons. Mutations were located in the SH2 domain of STAT3, which is a known mutational hotspot. No STAT5B mutations were found. In blood smears, overrepresentation of large granular lymphocytes was observed, and in the majority of cases the CD8+ T-cell receptor repertoire was skewed when analyzed by flow cytometry. In bone marrow biopsies, an increased amount of phospho-STAT3 positive cells was discovered. Plasma cytokine profiling showed that ten of the 92 assayed cytokines were elevated both in Felty syndrome and large granular lymphocyte leukemia, and three of these cytokines were also increased in patients with uncomplicated rheumatoid arthritis. In conclusion, somatic STAT3 mutations and STAT3 activation are as frequent in Felty syndrome as they are in large granular lymphocyte leukemia. Considering that the symptoms and treatment modalities are also similar, a unified reclassification of these two syndromes is warranted.

Precise vergence control in complex scenes

In binocular systems, vergence is the process of directing the gaze so that the optical axes intersect at the point of interest. Region based methods of disparity analysis provide fast and reliable estimates of the vergence error. Unfortunately, it is difficult to determine on what image features these approaches are in fact verging. Previous approaches to vergence control have for the most part failed to ensure that both cameras actually verge on the object of interest, especially in complex scenes. This paper presents a system that addresses this problem. By using the cepstral filter in a multiresolution setting with a dominant camera, our system can verge accurately in complex scenes. Specifically, the system adaptively refines the vergence angle in a scale space consisting of the center patches of a Gaussian pyramid. The effects of the cepstrum in a multiresolution system are analyzed, and the precision and performance of the new system are verified on natural scenes.

PRISM: a visual programming system for pipelined image processors

Pipelined image processing hardware has become increasingly popular because it makes it possible to build real-time machine vision systems at reasonable cost. Unfortunately, this type of hardware is often difficult to program, and the difficulty increases rapidly as the machines become more flexible and powerful. In this paper we present PRISM, a visual programming language that supports rapid prototyping and algorithm development on pipelined image processors. Computations are represented by graphs whose nodes are data transformations and whose arcs are data paths. The system allows the user to build and edit graphs and attach attributes to graph nodes specifying details of the computation (gains, masks, et cetera). Once the graph is adequately connected, the system traverses the graph, analyzes the data dependencies, and constructs an execution schedule. It then repeatedly executes the schedule, mapping graph nodes to specific hardware resources as needed. We discuss the overall architecture of the system, describe the class of hardware devices to which it is applicable, and then present an implementation for the Datacube MV 20. We analyze the implementation in terms of how well it makes use of the underlying hardware, and discuss ways of improving its efficiency.

Abstract 606: Novel somatic mutations in the DNA-binding and coiled-coil domain of the STAT3 gene in LGL-leukemia

Introduction T-cell large granular lymphocyte (T-LGL) leukemia is a rare, clonal disease characterized by the expansion of CD8+ cytotoxic T-cells. We recently discovered that 40% of T-LGL leukemia patients have somatic mutations in the SH2-domain of the STAT3 gene (Koskela et al. NEJM 2012). As aberrant STAT3 activation can be observed in all patients with LGL-leukemia, we now aimed to discover whether patients without mutations in the STAT3 hotspot area harbor mutations in the other parts of the STAT3 gene. Methods Targeted STAT3 sequencing covering all 23 coding exons was done with in-house developed deep amplicon sequencing panel using the Illumina Miseq platform. The data was analyzed with a bioinformatics pipeline, which is based on calling of variants with specific counts/frequencies and filtering out false positives using the estimated error rate and quality data of amplicon reads. All samples with a frequency ratio ≥0.9 were considered to be true mutations after filtering of SNPs and low coverage variants. 111 LGL-leukemia patients with no known STAT3-mutations in the SH2 domain were analyzed. To explore the functional effects of mutations, expression constructs were generated with the identified variants and wild-type STAT3. The variants were expressed in HEK-293 cells carrying a STAT3-responsive SIE-reporter driven luciferase expression sequence to establish increased basal and IL6-stimulated STAT3 activity. Results With targeted amplicon sequencing, 3 patients were discovered to have STAT3 missense mutations in the DNA-binding domain. Two patients presented with the same H410R mutation with a variant allele frequency (VAF) of 49% and 8.8% respectively while another patient had a S381Y mutation (VAF 7%). The mutation H410R occurs in the DNA-binding domain in a highly conserved position, and results in conversion of histidine to arginine, which would predict for a slight increase in hydrophilicity. In addition to STAT3 DNA-binding domain mutations, one T-LGL patient had a novel F174S mutation in the coiled-coil domain of STAT3 (VAF 43%). The coiled-coil domain of STAT3 has previously been shown to be essential in SH2-domain mediated receptor binding and subsequent activation. Luciferase measurements of SIE-reporter HEK-293 cells transfected with constructs expressing either wild-type, variant F174S, H410R or Y640F STAT3 (the most common activating mutation in LGL leukemia) revealed the F174S and H410R variants to be as activating as the Y640F mutation in both unstimulated and IL-6 stimulated conditions. Conclusions T-LGL leukemia patients without STAT3 SH2-domain mutations harbor novel activating mutations in the DNA-binding and coiled-coil domain of STAT3. The frequency of mutations was 3.6% (4 of 111 patients). These findings further highlight the importance of screening the whole STAT3 gene in the diagnostic workup of LGL-leukemia and the central role of STAT3 in the pathogenesis of the disease. Citation Format: Emma I. Andersson, Hanna Rajala, Heikki Kuusanmaki, Arjan van Adrichem, Samuli Eldfors, Sonja Lagstrom, Thomas Olson, Michael Clemente, Pekka Ellonen, Caroline Heckman, Thomas P. Loughran, Jaroslaw P. Maciejewski, Satu Mustjoki. Novel somatic mutations in the DNA-binding and coiled-coil domain of the STAT3 gene in LGL-leukemia. [abstract]. In: Proceedings of the 106th Annual Meeting of the American Association for Cancer Research; 2015 Apr 18-22; Philadelphia, PA. Philadelphia (PA): AACR; Cancer Res 2015;75(15 Suppl):Abstract nr 606. doi:10.1158/1538-7445.AM2015-606