<?xml version="1.0" encoding="UTF-8"?><rss xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:content="http://purl.org/rss/1.0/modules/content/" xmlns:atom="http://www.w3.org/2005/Atom" version="2.0" xmlns:cc="http://cyber.law.harvard.edu/rss/creativeCommonsRssModule.html">
    <channel>
        <title><![CDATA[Stories by Fernando Coelho on Medium]]></title>
        <description><![CDATA[Stories by Fernando Coelho on Medium]]></description>
        <link>https://medium.com/@omninando?source=rss-423ee9db5a5a------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/1*RVJVP9oqgRQy6JxEn9d9EQ.jpeg</url>
            <title>Stories by Fernando Coelho on Medium</title>
            <link>https://medium.com/@omninando?source=rss-423ee9db5a5a------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Tue, 17 Nov 2026 14:11:41 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@omninando/feed" rel="self" type="application/rss+xml"/>
        <webMaster><![CDATA[yourfriends@medium.com]]></webMaster>
        <atom:link href="http://medium.superfeedr.com/" rel="hub"/>
        <item>
            <title><![CDATA[Design System Documentation at Scale: Leveraging AI and Automation]]></title>
            <link>https://medium.com/typeforms-engineering-blog/design-system-documentation-at-scale-leveraging-ai-and-automation-b09c9e1a8921?source=rss-423ee9db5a5a------2</link>
            <guid isPermaLink="false">https://medium.com/p/b09c9e1a8921</guid>
            <category><![CDATA[notion]]></category>
            <category><![CDATA[docusaurus]]></category>
            <category><![CDATA[mcps]]></category>
            <category><![CDATA[ai]]></category>
            <category><![CDATA[design-systems]]></category>
            <dc:creator><![CDATA[Fernando Coelho]]></dc:creator>
            <pubDate>Fri, 03 Oct 2025 13:35:07 GMT</pubDate>
            <atom:updated>2025-10-03T13:35:07.904Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*t4ac6m-HjuydXJusIE4lcA.jpeg" /></figure><p>In today’s fast-paced development environment, maintaining up-to-date design system documentation while minimizing maintenance overhead presents a significant challenge. At Typeform, we’ve developed an innovative approach that leverages AI tools and Notion as a headless CMS, and automated workflows to ensure our documentation remains current with minimal manual intervention.</p><h3>The challenge</h3><p>Design systems are living entities that constantly evolve. As components, patterns, and guidelines change, documentation must keep pace. The biggest challenge was having information scattered across multiple platforms. We had Figma for design specs, Notion for guidelines, and Storybook for code examples, with frequent inconsistencies between them. Developers would see one implementation in Storybook that didn’t match what designers had documented in Figma, while separate guidelines existed in Notion documents. This fragmentation created confusion, slowed development, and led to implementation errors as team members struggled to identify the single source of truth for any given component or pattern.</p><p>Additionally, documentation often becomes siloed in design tools, making it inaccessible to developers who need it most. The distance between Figma documentation and developer-consumed resources creates a communication gap that slows development and introduces inconsistencies. Organizations also face high maintenance costs for custom documentation sites, while contribution friction discourages team members from making regular updates to keep documentation current.</p><h3>Our architecture: Monorepo approach</h3><p>We structured our design system as a monorepo containing both packages and applications.</p><h4>Packages</h4><p>We have 3 core packages in our design system: UI components which contains our React components. Design tokens, which contains our primitives for color, spacing and typography, and Assets which contains our icons, illustrations and other visual elements. These packages also encompass animations with standardized motion patterns and transitions, code snippet examples consumed by multiple applications, and a Notion API Client that fetches and transforms documentation content.</p><h4>Applications</h4><p>Within the monorepo, we maintain a documentation website built with Docusaurus for comprehensive documentation, a Sandbox environment serving as an interactive playground for component experimentation, and an MCP Server providing AI-powered assistance for engineers and designers.</p><h3>How the system works</h3><h4>Notion as a Content Management System</h4><p>We built a system that democratizes contribution to a bespoke implementation of our documentation website while simultaneously feeding critical design system resources to our Model Context Protocol server. At the core of this approach is Notion, which serves as an accessible content management system where anyone on the team can contribute documentation using a familiar interface. Since our company already extensively uses Notion on a daily basis, it was only natural to leverage it for our documentation needs. This approach lowered the barrier of entry significantly, as we already had the Notion API setup in place. The API enables us to programmatically fetch and transform this content into markdown, which then flows to multiple destinations within our monorepo.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*qGdr-DKEPj93DShWtbJXaw.png" /></figure><h4>Leveraging Notion databases for rich metadata</h4><p>The power of this approach became particularly evident through our use of Notion’s database page type. By structuring component documentation as database entries with rich metadata, we created a system that is both easy to update and remarkably information-dense. Each database row can contain API reference links, component status indicators, page thumbnails, and recursive database relationships that link related components and patterns. This structure allows us to generate comprehensive, interconnected documentation while maintaining it through a simple, familiar interface that requires no technical knowledge to update.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*s7D0GKIRXsTWHP3B9aDo5g.png" /></figure><h4>Documentation website with Docusaurus</h4><p>The documentation website itself is built with Docusaurus, providing the flexibility needed for a fully branded and customized experience. Despite requiring more maintenance than simpler solutions, Docusaurus gives us powerful MDX support for interactive documentation, built-in versioning to track design system evolution, and comprehensive search capabilities. The transformation pipeline converts Notion content into the markdown and MDX formats that Docusaurus consumes, handling everything from simple text blocks to complex code examples and interactive demos.</p><h3>Automated publishing pipeline</h3><p>This architecture enables the production website to be updated without writing a single line of code. A GitHub Actions automation runs weekly and can also be triggered manually, fetching updated content from Notion and opening a pull request with the latest changes. Each PR requires review from both an engineer and a designer, maintaining quality standards while allowing anyone on the team to contribute documentation updates through Notion alone.</p><h4>AI-Powered assistance via MCP</h4><p>The same content pipeline that feeds our documentation website also powers our Model Context Protocol server. The MCP server consumes multiple resources from the monorepo: the markdown documentation generated from the Notion API, our library of design tokens containing variables for colors, spacing, and typography, and the shared code snippets that demonstrate real-world usage patterns. This unified approach means the AI-powered assistant has access to the same documentation, tokens, and examples that developers and designers use, enabling it to provide contextual guidance, suggest appropriate components based on requirements, and highlight accessibility considerations during development.</p><h3>Shared code snippets</h3><p>A critical aspect of our documentation strategy is a package that maintains a single source of truth for code snippets. By consuming this package across the documentation site, sandbox environment, and MCP server, we ensure consistency across all platforms while providing realistic implementation patterns for developers and training data for the AI-powered assistant. To illustrate that, our component pages feature an interactive playground with editable code examples that can be modified in real-time and seamlessly transferred to our sandbox for further experimentation, debugging, and implementation of new features in an isolated environment.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*RpJ8-P-c1TM780xV9T0j6g.gif" /></figure><h3>Benefits realized</h3><p>This architecture has delivered several key advantages across the organization. Designers benefit from a simplified contribution process through the familiar Notion interface, spending less time maintaining documentation and more time on design work. Developers gain access to always-current documentation through multiple channels, with AI-assisted implementation reducing development time and consistent examples demonstrating best practices. The organization as a whole experiences higher adoption rates of design system components, improved product consistency, reduced design debt through better documentation, and lower maintenance costs despite the sophisticated documentation system.</p><h3>Challenges and lessons learned</h3><h4>Technical hurdles</h4><p>While our approach has been largely successful, we encountered several technical challenges. We needed to adapt how Notion routes between pages to match Docusaurus’s static site routing structure to maintain parity and minimize changes to how team members author content in Notion. Additionally, we implemented a careful fetching strategy to avoid being rate-limited by Notion’s API infrastructure.</p><h4>Documentation segmentation</h4><p>Another challenge was determining which content to write in markdown versus which to render as React components for specialized visuals. Styled markdown has limitations when presenting complex interactive examples and custom UI patterns. For instance, our design tokens — while displayed in table format — required rendering complex elements after being consumed by the tokens package.</p><h4>Future improvements</h4><p>We continue to refine our documentation pipeline with plans to implement real-time updates for critical documentation changes and expand MCP capabilities to include code generation from design specs. We’re also looking at adding usage analytics to identify documentation gaps and introduce interactive tutorials generated from documentation content.</p><p>By combining Notion’s user-friendly interface, automated transformation pipelines, and AI-powered assistance, we’ve created a documentation system that scales with our design system while minimizing overhead maintenance. By creating a custom, streamlined design system documentation solution, we’ve already seen a big uptick in UI consistency and less back and forth between designers and engineers.</p><p>Design system documentation is always a challenge to keep up to date and relevant. By investing in a, on one hand custom, but at the same time automated solution where anyone can contribute all whilst baking in AI and MCP functionality, we can ensure Echo is truly setup for the future.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=b09c9e1a8921" width="1" height="1" alt=""><hr><p><a href="https://medium.com/typeforms-engineering-blog/design-system-documentation-at-scale-leveraging-ai-and-automation-b09c9e1a8921">Design System Documentation at Scale: Leveraging AI and Automation</a> was originally published in <a href="https://medium.com/typeforms-engineering-blog">Typeform&#39;s RnD Blog</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Alem das Cores]]></title>
            <link>https://medium.com/alem-das-cores/al%C3%A9m-das-cores-40b371520df7?source=rss-423ee9db5a5a------2</link>
            <guid isPermaLink="false">https://medium.com/p/40b371520df7</guid>
            <category><![CDATA[color-blindness]]></category>
            <category><![CDATA[daltonismo]]></category>
            <dc:creator><![CDATA[Fernando Coelho]]></dc:creator>
            <pubDate>Sat, 27 Jun 2015 19:03:35 GMT</pubDate>
            <atom:updated>2016-03-29T19:10:56.443Z</atom:updated>
            <content:encoded><![CDATA[<h2>Além das Cores</h2><h4>Uma perspectiva um tanto diferente do mundo</h4><p>É interessante ver a reação das pessoas quando digo que sou daltônico. Eu já estou acostumado com perguntas como &quot;que cor é essa?&quot;, &quot;como você consegue ver o semáforo?&quot;, ou &quot;você consegue ver {{insira a cor}}?&quot;. O mais interessante é como parece tão alienígena para as pessoas, saberem que eu não tenho a mesma percepção que elas sobre o mundo que vêem. O espanto é ainda maior quando digo que estudo/trabalho com Design. Afinal, como pode alguém que não vê cores trabalhar com algo que depende tanto destas?</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/550/1*LCObsyjVIXAKC6CWXvNGhQ.png" /><figcaption>“Problemas de daltonico: Pessoas perguntando que cor é essa; Jogos baseados em cores; Semáforos.” Tenso.</figcaption></figure><p>Conversando muito e procurando nas interwebs sobre o assunto, percebi que não é um assunto de conhecimento geral e a maior parte das pessoas não sabe muito bem o que significa ou no que implica ser daltônico. De fato existem várias fontes sobre Daltonismo ou mesmo Design acessível para daltônicos, porém, infelizmente não é muito comum este tipo de conteúdo em português. Quero, por meio deste e outros posts, tentar disseminar mais sobre acessibilidade como um todo, já que não é apenas uma questão de adequação e sim de vantagem competitiva.</p><h4>O mundo como eu vejo</h4><p>Primeiro de tudo, <strong>não </strong>é como ver em preto e branco ou escalas de cinza, entretanto, apesar de raros, existem casos de pessoas não vêem cor alguma. Eu consigo ver a maior parte das cores, porém apenas individualmente. O problema se dá principalmente em agrupamento de cores, ainda mais se essas possuem o contraste semelhante. O tipo de Daltonismo que possuo é a <a href="https://en.wikipedia.org/wiki/Color_blindness#Classification"><strong><em>Deuteranomalia</em></strong></a>, a mais comum, nos quais os tons de verde e vermelho não conseguem ser diferenciados pelos receptores da retina.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/690/1*13f6vujdzhcdOBd8nCrAZA.jpeg" /><figcaption>Yup, as duas primeiras fotos são iguais pra mim. Essa imagem veio <a href="http://wearecolorblind.com/article/a-quick-introduction-to-color-blindness/">daqui</a>.</figcaption></figure><p>Mas como eu sei que sou daltônico? É simples. Você com certeza já viu em algum lugar, aqueles quadros com várias bolinhas coloridas com um numeral formado entre estas bolinhas. É o <a href="https://en.wikipedia.org/wiki/Ishihara_Test">Teste de Ishihara</a> no qual consiste em:</p><blockquote>“uma série de cartões <a href="https://pt.wikipedia.org/wiki/Cor">coloridos</a>, cada um contendo vários círculos feitos de cores ligeiramente diferentes das cores daqueles situados nas proximidades. Seguindo o mesmo padrão, alguns círculos estão agrupados no meio do cartão de forma a exibir um número que somente será visível pelas pessoas que possuirem <a href="https://pt.wikipedia.org/wiki/Vis%C3%A3o">visão</a> normal.”</blockquote><figure><img alt="" src="https://cdn-images-1.medium.com/max/600/1*qfTlR1qIqNlKUCc4-TMYAg.png" /><figcaption>Pessoas com visão normal devem ver claramente o numeral “74”. Indivíduos com algum nivel de daltonismo podem ver o numero “21” ou podem não ver numero algum. Essa foto veio <a href="https://en.wikipedia.org/wiki/Color_blindness#Classification">daqui</a>.</figcaption></figure><h4>Mas o que isso significa?</h4><p>Muita coisa. Significa que tudo que consumo e faço uso visualmente, seja na TV, impresso ou online é afetado. Quando o que vejo é baseado unicamente em cores, perde-se grande parte ou todo do sentido pra mim. Se as combinações de cores não forem feitas com muito cuidado a mensagem não é transmitida com clareza, ou não transmitida <em>at all</em>.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/604/1*TbGwyKiWLZi0baZAvLYw9Q.jpeg" /><figcaption>Eu, sempre que vejo algo baseado somente em cores. “Não tenho certeza se erro ou não sabe a verdade, porque daltônico”</figcaption></figure><h4>Como ver sem as cores</h4><p>Quando queremos consumir conteúdo ou fazer uso de uma ferramenta nós daltônicos precisamos de algo natural e não intrusivo que auxilie o uso das cores, como símbolos e padrões. Se isso não for possível, é necessário utilizar cores com contrastes diferentes. Para ilustrar o que quero dizer, temos o excelente trabalho do <a href="http://www.trello.com/">Trello</a>, ferramenta de organização de tarefas, que possui modo daltônico e exibe, entre outras coisas, padrões auxiliando as cores das labels.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*kEsGtr1PmqpDAxOf2hPmSQ.png" /><figcaption>Solução do Trello para tornar as labels acessíveis a daltonicos</figcaption></figure><p>Outro exemplo magnífico é <a href="http://weplaydots.com/twodots.html">Two Dots</a>, um jogo focado em fazer conexões com pontos de mesma cor que também tem um modo daltônico. Sou fã incondicional, especialmente por seu esmero com Design e Acessibilidade.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*NsZ387vyvftsWHv8oqTNOA.png" /><figcaption>O Two Dots auxilia a identificaçao dos pontos com simbolos</figcaption></figure><h4>Bom, mas porque isso tudo importa?</h4><p>Porque quero fazer do mundo um lugar mais acessível e fazer as pessoas conhecerem mais sobre esse assunto é um ótimo início. Mesmo que pessoas com daltonismo e outros tipos de limitação representem uma pequena parcela da população, é fundamental pensar em acessibilidade como forma de inclusão e, porque não, vantagem competitiva. Se tivermos essa concepção em todos os nossos projetos, será algo natural e não aquela obrigação de cumprir tabela. Faremos mais coisas em que todos possam usufruir igualmente, resultando numa experiência de usuário muito melhor.</p><p>Vou explorar melhor as facetas do daltonismo em outros posts além de comEu compartilharei mais sobre minhas experiências como daltônico e outros assuntos <a href="http://medium.com/al%C3%A9m-das-cores">aqui</a>. Quer entrar em contato? Fale no <a href="http://facebook.com/nandoacoelho">Feice</a> ou mande um <a href="http://twitter.com/nandoacoelho">tuíte</a>!</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=40b371520df7" width="1" height="1" alt=""><hr><p><a href="https://medium.com/alem-das-cores/al%C3%A9m-das-cores-40b371520df7">Alem das Cores</a> was originally published in <a href="https://medium.com/alem-das-cores">Alem das Cores</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Sickle cell identification in Red Blood Cell Images]]></title>
            <link>https://medium.com/@omninando/sickle-cell-identification-in-red-blood-cell-images-4ff6ff1d9368?source=rss-423ee9db5a5a------2</link>
            <guid isPermaLink="false">https://medium.com/p/4ff6ff1d9368</guid>
            <dc:creator><![CDATA[Fernando Coelho]]></dc:creator>
            <pubDate>Fri, 19 Dec 2014 08:16:56 GMT</pubDate>
            <atom:updated>2014-12-19T08:18:33.646Z</atom:updated>
            <content:encoded><![CDATA[<p><strong>Ana Luiza Motta Gomes and Fernando de Almeida Coelho</strong><br>Mathematics and Computer Science Department, Clark University, Worcester, United States of America</p><blockquote><strong><em>Abstract — </em>Sickle cell disease is a blood disorder that affects hemoglobin, the molecule in red blood cells that delivers oxygen to cells throughout the body. People with this disorder produce an atypical hemoglobin molecule that makes the body produce sickle-shaped cells. Even though the detection and counting of these cells is very important to start the process of treatment of many diseases as soon as possible, scientists still do this process manually, which is very time-consuming. In this paper, it is presented an approach to design a system that can detect and count sickle cells automatically using methods such as image segmentation and ellipse fitting.</strong></blockquote><h4>I. INTRODUCTION</h4><p>Sickle cell disease is a major public health concern that affects millions of people around the world. In United States, it is estimated that sickle cell disease affects nine to a hundred thousand Americans [1]. In addition, more than three million people in the United States carry the gene that allows them to potentially pass the disease onto their children. The disease is an inherited blood disorder that costs the red blood cells (erythrocytes) to be abnormally sickle-shaped for carrying a different type of hemoglobin. Because of this abnormal shape, they don’t carry oxygen for very long, they don’t live as long as normal red blood cells and people who has this disease end up with organ damage over time. The first symptom of the sickle cell disease is anemia. People also tend to have painful episodes, which are unexpectable, unpredictable and can be of different levels of severity.</p><p>In order to confirm the sickle cell disease diagnose, a sample of blood is analyzed to check for the presence of the sickle-shaped erythrocytes and compare their frequency to normal red blood cells. Scientists tend to identify and count these cells manually, using a hand counter most of the time. The identification depends on the evaluator’s ability and knowledge, being affected by presence of different layers of cell and their position. This situation leads to a tedious and time-consuming work as well as an inaccuracy on the count that can be improved by automated methods. Blood sample images taken from microscope can provide important qualitative and quantitative information. Only a few attempts of partial / full automated systems based on image processing are presented in literature and most of them are in prototype stage [2].</p><p>As technology becomes more widespread into the biomedical area, researchers are making big efforts to find a way to identificate different hematic pathologies and count red blood cells automatically. Computer Vision, which is a subarea of Computer Science that can make computers understand the visual world of humans, is able to provide useful techniques to accomplish this task.</p><p>This paper focuses on the identification of the shape of sickle cells among red blood cells in microscopic images. The proposed work is to build an automated system for the detection of sickle cells among red blood cells by using Conic Fitting method [3] and subsequently count red blood cells in these images. The implementation of this system uses OpenCV [4] library with Java interface.</p><h4>II. METHODS</h4><p><em>A. Pre-processing — </em>For the purpose of the work to be achieved, various pre- processing techniques were applied to the microscopic images, such as converting them to gray scale, spatial smoothing with Gaussian Blur [5] and Adaptive Thresholding in addition to morphological transformations if necessary [6].</p><p>The first pre-processing technique used on this work was to convert all the sample images to gray scale, in order to eliminate the Hue and Saturation levels while retaining luminance. Converting images to gray scale makes the convolution to be used, what is less computationally expensive. After that, Gaussian blur is applied to make the image smoother using a convolution kernel. With more smoothing, fewer edges are detected, which means that only the edges of the cells will be highlighted.</p><p>Adaptive Thresholding is used to transform the resulting images into binary images. As the color of cells can slightly change, an adaptive value for thresholding was chosen using Gaussian methods, which can change by taking into account the pixel neighbors when iterating through the image. Morphological operations, such as dilation and/or erosion can be applied if the resulting binary images are not satisfactory in highlighting the edges.</p><p><em>B. Finding Contours — </em>To accomplish the task of finding the contours of each red blood cell, the Border Following algorithm [7] was used. This algorithm is responsible for extracting the topological structure of a given binary image. In practice, the algorithm discriminates between outer and inner borders and classifies them hierarchically. In this work, only the outermost borders of the cells found in the samples were used.</p><p><em>C. Ellipse Fitting — </em>An ellipse is a conic section and its shape can be represented by an eccentricity, which is a measure that ranges from 0 to 1. If the value is very close to 0, the ellipse has a circular-like shape. In the other hand, if the ellipse has a high eccentricity, its shape is more likely to be a flattened circle. According to that, both sickle and regular red blood cells can be considered as ellipses. The algorithm for fitting ellipses proposed by Fitzgibbon, A.W. <em>et al. </em>[3], showed to be very useful if applied to this work.</p><p><em>D. Counting — </em>Given the rectangle found previously by the Ellipse Fitting method, two assumptions were made. Sickle cells usually have an elongated form, thus making them thinner than a regular or even a shifted red blood cell. Taking that in consideration, if the width of the bounding rectangle is between 20% and 50% of the height is more likely to be a sickle cell. In contrast, if the width is more than 50% of the height and between a determined threshold for the size, it is more likely to be a regular blood cell.</p><h4>III. RESULTS</h4><p><em>A. Experimental Results — </em>For this study, five blood cell images were used. Each of the images were pre-processed by the above mentioned techniques. Finally, the number of sickle and regular red blood cells were identified and counted in each and every image finding the contours and fitting ellipses.</p><p>The experimental result of one of the images is shown below:</p><p><em>B. Discussion — </em>Results showed a good performance in detecting both sickle-shaped and circular-shaped cells. However, the system detected some false positives for each of the shapes, which means that the detection of sickle cells is not completely accurate.</p><p>For the purpose of counting red blood cells, the system worked well by identifying most of the cells, but the partial ones in the edges of the images were still difficult to identify. The table below shows the relation between cells counted manually and cells counted methodically.</p><h4>IV. CONCLUSION</h4><p>This paper presents a methodology to automate the identification and counting of sickle and red blood cells using the contours and ellipse fitting. Results indicate that the identification of sickle cells have a remarkable accuracy. It is also very time-effective as manual counting is a very time consuming and tedious job. However, the software must be improved to count more effectively the red blood cells by identifying overlapping cells and the partial ones in the edges of the images. Also automate the detection of the magnification level by identifying the average size of the cells.</p><p>Further studies will be focused on the perfect detection of sickle cells, as well as the complete cell count, i. e. total count of red blood cells, white blood cells and different types of abnormalia.</p><h4>REFERENCES</h4><p>[1] “Sickle Cell Disease (SCD): Data &amp; Statistics.” Centers for Desease Control and Prevention. <a href="http://www.cdc.gov/ncbddd/sicklecell/data.html">http://www.cdc.gov/ncbddd/sicklecell/data.html</a> Acessed, December 12, 2014.</p><p>[2] M. Mausumi, R. K. Gupta, and M. Mukherjee. “Detection and counting of red blood cells in blood cell images using Hough transform.” International Journal of Computer Applications 53, no. 16, pp. 18–22, 2012.</p><p>[3] A. W. Fitzgibbon, M. Pilu, and R. B. Fisher. “Direct least square fitting of ellipses.” Pattern Analysis and Machine Intelligence, IEEE Transactions on 21.5, pp. 475–480, 1999.</p><p>[4] Itseez “OpenCV” OpenCV.org. <a href="http://www.opencv.org/">http://www.opencv.org/</a> Accessed December 13, 2014.</p><p>[5] R. Szeliski. “Computer vision: algorithms and applications.” Springer, pp. 114–19, 2010.</p><p>[6] Ibid. 130–132.</p><p>[7] S. Suzuki, “Topological structural analysis of digitized binary images by border following.”, Computer Vision, Graphics, and Image Processing 30, no. 1, pp. 32–46, 19.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=4ff6ff1d9368" width="1" height="1" alt="">]]></content:encoded>
        </item>
    </channel>
</rss>