matte-esse.github.io/index.html at main · matte-esse/matte-esse.github.io · GitHub

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
<!DOCTYPE html>
<html lang="en">
<head>
    <meta charset="UTF-8">
    <meta name="viewport" content="width=device-width, initial-scale=1.0">
    <title>THRIVE</title>
    <link rel="stylesheet" href="style.css">
    <link rel="preconnect" href="https://fonts.googleapis.com">
    <link rel="preconnect" href="https://fonts.gstatic.com" crossorigin>
    <link href="https://fonts.googleapis.com/css2?family=Raleway:wght@700&family=Source+Serif+Pro:ital,wght@0,400;0,700;1,400&family=Open+Sans:wght@400;600&display=swap" rel="stylesheet">
</head>
<body>
    <header>
        <div class="top-bar"></div>
    </header>

    <main>
        <h1 class="title">T H R I V E</h1>

        <h2 class="subtitle">
            <span class="big-letter">T</span>hymoma
            <span class="big-letter">H</span>istology
            <span class="big-letter">R</span>ecognition
            <span class="big-letter">I</span>nitiative for
            <span class="big-letter">V</span>alidation and
            <span class="big-letter">E</span>xpansion
        </h2>

        <p class="description">
            THRIVE aims at collecting data to build advanced Deep Learning algorithms capable of classifying Thymic Malignancy subtypes (A, AB, B1, B2, B3, TC). Thymic malignancy are rare and DL model perform best when they are trained on large dataset. Furthermore, digital pathology based models are prone to batch effect &nbsp;<a href="#ref1" class="citation">[1]</a>, hence it's imperative to have datasets from multiple institution to train model that can learn biologically sound features and generalize well in the clinical setting.
        </p>

        <p class="contact">
            If you'd like to contribute please contact us: <a href="mailto:mattsacco@uchicago.edu">mattsacco@uchicago.edu</a>
        </p>

        <hr class="divider">

        <div class="columns">
            <div class="column">
                <h3>Cite us</h3>
                <p>
                    Cite our first abstract <i>Automated histologic subtyping of thymic epithelial tumors with deep learning</i> &nbsp;<a href="#ref2" class="citation">[2]</a> where we proved Deep Learning can learn features associated with TETs subtypes. Our second abstract in which we trained a model that is robust to an external validation set is coming soon.
                </p>
            </div>

            <div class="column">
                <h3>Try our tools</h3>
                <p>
                    Try our tool <i><a href="https://github.com/slideflow/slideflow" target="_blank">Slideflow</a></i> &nbsp;<a href="#ref3" class="citation">[3]</a>. If your interested in conducting your own digital pathology research, check out our open source code base <i>Slideflow</i>.
                </p>
            </div>
        </div>

        <div class="references">
            <p id="ref1">
                [1] Howard, F.M., Dolezal, J., Kochanny, S. et al. <a href="https://doi.org/10.1038/s41467-021-24698-1" class="ref-link" target="_blank"><i>The impact of site-specific digital histology signatures on deep learning model accuracy and bias.</i> Nature Communications (2021)</a>
            </p>
            <p id="ref2">
                [2] Dolezal JM, Guo W, Bestvina C, Vokes E, Donington J, Husain A, Garassino MC. <a href="https://pmc.ncbi.nlm.nih.gov/articles/PMC9792820/" class="ref-link" target="_blank"><i>Automated histologic subtyping of thymic epithelial tumors with deep learning.</i> Mediastinum (2022)</a>
            </p>
            <p id="ref3">
                [3] Dolezal, J.M., Kochanny, S., Dyer, E. et al. <a href="https://doi.org/10.1186/s12859-024-05758-x" class="ref-link" target="_blank"><i>Slideflow: deep learning for digital histopathology with real-time whole-slide visualization.</i> BMC Bioinformatics (2024)</a>
            </p>
        </div>
    </main>
</body>
</html>