Development and Evaluation of interest Point Detection for Neural Imaging

نویسندگان

  • CHRISTER GUSTAVSSON
  • Danica Kragic
چکیده

In 1998 a new method to reduce the production of an optional protein in a cell was discovered, which enabled a new way of performing high throughput experiments. The effect of individual proteins on essential cell-properties could be analyzed by silencing one protein at a time. These experiments resulted in large quantities of data that had to be analyzed. In practice this was only possible if automated computer algorithms were available. The human neural system consists of more than one hundred billion nerve cells (called neurons) working together in a complex network enabling us to walk, eat, understand and think among other things. The healthiness of a neuron is affected by the access of different proteins during the growth and can be measured by counting the average number of synapses per area in the branches of the neuron. The process of evaluating the effect of knocking out proteins during the development can be rendered more effective if automated computer algorithms are utilized. This thesis is focused on developing and evaluating tools that can be used to study neuron development. The synapses can be characterized as intensity peaks with a radius of 2-3 pixels located on an unknown background. Effective algorithms to find these intensity peaks had to be developed to make the required measurements possible. Four different methods to find synapses in neural images are described and evaluated in this thesis. Methods based in the frequency domain prove to be unstable and lead to numerous of false positives. A spatial method, measuring the relation of the spread and the depth of the intensity peaks simultaneously, put all the other approaches in the shadow. The reliability of detection algorithms are heavily affected by the quality of the input images. To achieve equivalent inputs several pre-processing methods are discussed. Background subtraction methods are used to set the background to zero and contrast stretching methods are used to match the foregrounds of the inputs. By making the input images more equivalent the crucial detection variables can be fixed without deteriorating the result markedly. Utveckling och utvärdering av detektionsalgoritmer för neural bildbehandling Sammanfattning Under 1998 upptäcktes en ny metod för att dämpa produktionen av ett valfritt protein i en cell. Detta möjliggjorde ett nytt sätt att utföra kvantitativa cellexperiment. Inverkan av enskilda proteiner på viktiga cellegenskaper kunde analyseras genom att hindra produktionen av ett protein i taget. Denna form av experiment resulterar i stora kvantiteter data som måste analyseras. I praktiken kan detta endast genomföras om automatiska datoralgoritmer finns tillgängliga. Det mänskliga nervsystemet består av över hundra miljarder nervceller. Dessa samarbetar i ett komplext nätverk, vilket gör det möjligt för oss att gå, äta, tänka och minnas som några få exempel. En neurons hälsa påverkas av tillgången av olika proteiner under uppväxten och kan mätas genom att räkna antalet synapser per area i neuronens armar. Arbetet med att utvärdera effekten av att blockera protein i utvecklingsfasen kan effektiviseras genom att använda automatiska dataalgoritmer. Detta examensarbete är fokuserat på att utveckla och utvärdera verktyg som kan användas för att studera en neurons utveckling. Synapserna kan kännetecknas som intensitetstoppar med 2-3 pixlars radie placerade i en okänd bakgrund. Effektiva algoritmer för att detektera dessa intensitetstoppar var nödvändiga för att kunna utföra mätningarna. Fyra olika metoder för att detektera synapser i bilder tagna av neuroner är beskrivna och utvärderade i detta examensarbete. Frekvensbaserade metoder visar sig opålitliga med en stor frekvens av falska detektioner. En metod baserad på en mätning av intensitetstoppens höjd i relation till dess utbredning visade sig utklassa alla andra angreppssätt. Tillförlitligheten hos detektionsalgoritmerna är kraftigt påverkbara av bildkvaliteten hos inputbilderna. För att dessa bilder ska vara likvärdiga är förbehandling nödvändigt. Metoder att sätta bakgrunden till noll och skapa matchande kontrast hos förgrunden beskrivs. Genom att skapa likvärdiga inputbilder kan nödvändiga variabler i detektionsalgoritmerna sättas till statiska värden utan att försämra resultaten märkbart.

برای دانلود رایگان متن کامل این مقاله و بیش از 32 میلیون مقاله دیگر ابتدا ثبت نام کنید

ثبت نام

اگر عضو سایت هستید لطفا وارد حساب کاربری خود شوید

منابع مشابه

Detection of Alzheimer\\\\\\\'s Disease using Multitracer Positron Emission Tomography Imaging

Alzheimer's disease is characterized by impaired glucose metabolism and demonstration of amyloid plaques. Individual positron emission tomography tracers may reveal specific signs of pathology that is not readily apparent on inspection of another one. Combination of multitracer positron emission tomography imaging  yields promising results. In this paper, 57 Alzheimer's disease neuroimaging ini...

متن کامل

A review of neural network detection methods for breast cancer: review article

Breast cancer is the most common cancer among women and the earlier it is diagnosed, the easier it is to treat. The most common way to diagnose breast cancer is mammography. Mammography is a simple chest x-ray and a tool for early detection of non-palpable breast cancers and tumors. However, due to some limitations of this method such as low sensitivity especially in dense breasts, other method...

متن کامل

Neural Imaging Using Single-Photon Avalanche Diodes

Introduction: This paper analyses the ability of single-photon avalanche diodes (SPADs) for neural imaging. The current trend in the production of SPADs moves toward the minimumdark count rate (DCR) and maximum photon detection probability (PDP). Moreover, the jitter response which is the main measurement characteristic for the timing uncertainty is progressing. Methods: The neural imaging pro...

متن کامل

Conditional Random Fields for Airborne Lidar Point Cloud Classification in Urban Area

Over the past decades, urban growth has been known as a worldwide phenomenon that includes widening process and expanding pattern. While the cities are changing rapidly, their quantitative analysis as well as decision making in urban planning can benefit from two-dimensional (2D) and three-dimensional (3D) digital models. The recent developments in imaging and non-imaging sensor technologies, s...

متن کامل

طراحی یک مدل مبتنی بر شبکه‌های عصبی برای شناسایی و تجزیه و تحلیل الگوهای غیرطبیعی در نمودارهای کنترل فرآیند

Neural networks because of their abilities are used to patterns recognition. In statistical process control charts, a common cause variation distort expected form of unnatural patterns and so detection of assignable causes efficiently and precisely in a real-time is difficult. Therefore it would be logical to propose models based neural networks for recognition and analysis of patterns in proce...

متن کامل

Neural Network Based Protection of Software Defined Network Controller against Distributed Denial of Service Attacks

Software Defined Network (SDN) is a new architecture for network management and its main concept is centralizing network management in the network control level that has an overview of the network and determines the forwarding rules for switches and routers (the data level). Although this centralized control is the main advantage of SDN, it is also a single point of failure. If this main contro...

متن کامل

ذخیره در منابع من


  با ذخیره ی این منبع در منابع من، دسترسی به آن را برای استفاده های بعدی آسان تر کنید

عنوان ژورنال:

دوره   شماره 

صفحات  -

تاریخ انتشار 2005