<?xml version="1.0"?>
<oembed><version>1.0</version><provider_name>Grupo de Aplica&#xE7;&#xF5;es em Intelig&#xEA;ncia Artificial</provider_name><provider_url>https://wp.ufpel.edu.br/gaia</provider_url><author_name>gaia</author_name><author_url>https://wp.ufpel.edu.br/gaia/author/gaia/</author_url><title>Liver Tumor Segmentation in CT Scans Using Deep Learning</title><type>rich</type><width>600</width><height>338</height><html>&lt;blockquote class="wp-embedded-content" data-secret="UcWpBy3hUK"&gt;&lt;a href="https://wp.ufpel.edu.br/gaia/ieramiars1/"&gt;Liver Tumor Segmentation in CT Scans Using Deep Learning: A U-Net Approach with Transfer Learning&lt;/a&gt;&lt;/blockquote&gt;&lt;iframe sandbox="allow-scripts" security="restricted" src="https://wp.ufpel.edu.br/gaia/ieramiars1/embed/#?secret=UcWpBy3hUK" width="600" height="338" title="&#x201C;Liver Tumor Segmentation in CT Scans Using Deep Learning: A U-Net Approach with Transfer Learning&#x201D; &#x2014; Grupo de Aplica&#xE7;&#xF5;es em Intelig&#xEA;ncia Artificial" data-secret="UcWpBy3hUK" frameborder="0" marginwidth="0" marginheight="0" scrolling="no" class="wp-embedded-content"&gt;&lt;/iframe&gt;&lt;script type="text/javascript"&gt;
/* &lt;![CDATA[ */
/*! This file is auto-generated */
!function(d,l){"use strict";l.querySelector&amp;&amp;d.addEventListener&amp;&amp;"undefined"!=typeof URL&amp;&amp;(d.wp=d.wp||{},d.wp.receiveEmbedMessage||(d.wp.receiveEmbedMessage=function(e){var t=e.data;if((t||t.secret||t.message||t.value)&amp;&amp;!/[^a-zA-Z0-9]/.test(t.secret)){for(var s,r,n,a=l.querySelectorAll('iframe[data-secret="'+t.secret+'"]'),o=l.querySelectorAll('blockquote[data-secret="'+t.secret+'"]'),c=new RegExp("^https?:$","i"),i=0;i&lt;o.length;i++)o[i].style.display="none";for(i=0;i&lt;a.length;i++)s=a[i],e.source===s.contentWindow&amp;&amp;(s.removeAttribute("style"),"height"===t.message?(1e3&lt;(r=parseInt(t.value,10))?r=1e3:~~r&lt;200&amp;&amp;(r=200),s.height=r):"link"===t.message&amp;&amp;(r=new URL(s.getAttribute("src")),n=new URL(t.value),c.test(n.protocol))&amp;&amp;n.host===r.host&amp;&amp;l.activeElement===s&amp;&amp;(d.top.location.href=t.value))}},d.addEventListener("message",d.wp.receiveEmbedMessage,!1),l.addEventListener("DOMContentLoaded",function(){for(var e,t,s=l.querySelectorAll("iframe.wp-embedded-content"),r=0;r&lt;s.length;r++)(t=(e=s[r]).getAttribute("data-secret"))||(t=Math.random().toString(36).substring(2,12),e.src+="#?secret="+t,e.setAttribute("data-secret",t)),e.contentWindow.postMessage({message:"ready",secret:t},"*")},!1)))}(window,document);
//# sourceURL=https://wp.ufpel.edu.br/gaia/wp-includes/js/wp-embed.min.js
/* ]]&gt; */
&lt;/script&gt;
</html><thumbnail_url>https://wp.ufpel.edu.br/gaia/files/2025/12/eramia-logo-horizontal-1.png</thumbnail_url><thumbnail_width>612</thumbnail_width><thumbnail_height>200</thumbnail_height><description>Este trabalho apresenta um pipeline para segmenta&#xE7;&#xE3;o 2D de f&#xED;gado e tumores a partir de tomografia computadorizada (NIfTI). Utilizamos o dataset p&#xFA;blico Liver Tumor Segmentation com 131 volumes, do qual foram extra&#xED;das fatias 2D e convertidas em imagens JPEG (com compress&#xE3;o m&#xED;nima, sem perda percept&#xED;vel) para treinamento da U-Net com backbone ResNet50 (transfer learning) implementando a biblioteca FastAI. O pr&#xE9;-processamento inclui leitura dos volumes com NiBabel, aplica&#xE7;&#xE3;o de janelas DICOM espec&#xED;ficas para f&#xED;gado, normaliza&#xE7;&#xE3;o e redimensionamento para 128&#xD7;128. Como fun&#xE7;&#xE3;o de perda foi adotada CrossEntropy (multiclasse) e m&#xE9;tricas customizadas de acur&#xE1;cia sobre os pixels de interesse (foreground).</description></oembed>
