{
  "id": 170680,
  "title": "Pixel Histogram as a Feature",
  "url": "/competitions/osic-pulmonary-fibrosis-progression/discussion/170680",
  "author_name": "",
  "post_date": "2020-07-28T14:48:43.992743400Z",
  "votes": 4,
  "comment_count": 4,
  "views": 0,
  "content": "<p>\"AI-driven Detection, Characterization and Classification\nof Chronic Lung Diseases\"</p>\n\n<p>The whole thesis is probably quite useful.</p>\n\n<p>In particular I'm looking at an interesting idea on page 45 showing that there is a difference in the histograms of healthy and diseased lungs. Potentially a nice handcrafted feature. I'll see if I can come up with a notebook demonstrating it unless someone beats me to it.</p>",
  "messages": [
    {
      "id": "949309",
      "postDate": "07/28/2020 14:48:43",
      "content": "<p>\"AI-driven Detection, Characterization and Classification\nof Chronic Lung Diseases\"</p>\n\n<p>The whole thesis is probably quite useful.</p>\n\n<p>In particular I'm looking at an interesting idea on page 45 showing that there is a difference in the histograms of healthy and diseased lungs. Potentially a nice handcrafted feature. I'll see if I can come up with a notebook demonstrating it unless someone beats me to it.</p>",
      "rawMarkdown": "\"AI-driven Detection, Characterization and Classification\nof Chronic Lung Diseases\"\n\nThe whole thesis is probably quite useful.\n\nIn particular I'm looking at an interesting idea on page 45 showing that there is a difference in the histograms of healthy and diseased lungs. Potentially a nice handcrafted feature. I'll see if I can come up with a notebook demonstrating it unless someone beats me to it.",
      "votes": null
    },
    {
      "id": "949405",
      "postDate": "07/28/2020 16:11:01",
      "content": "<p>Thanks for posting the thesis! This definitely seems worth taking a look.</p>\n\n<p>I hope someone comes up with a real use-case of the histogram's kurtosis and skewness in their models.</p>",
      "rawMarkdown": "Thanks for posting the thesis! This definitely seems worth taking a look.\n\nI hope someone comes up with a real use-case of the histogram's kurtosis and skewness in their models.",
      "votes": null
    },
    {
      "id": "950108",
      "postDate": "07/29/2020 07:28:19",
      "content": "<p>It was also tackled <a href=\"https://www.kaggle.com/c/osic-pulmonary-fibrosis-progression/discussion/165727\">here</a>\nIt looks promising. It's on my todo list</p>",
      "rawMarkdown": "It was also tackled [here](https://www.kaggle.com/c/osic-pulmonary-fibrosis-progression/discussion/165727)\nIt looks promising. It's on my todo list",
      "votes": null
    },
    {
      "id": "950193",
      "postDate": "07/29/2020 08:41:58",
      "content": "<p>Ah thanks I'd managed to skip over that section! Well at least it's backed up twice now!</p>",
      "rawMarkdown": "Ah thanks I'd managed to skip over that section! Well at least it's backed up twice now!",
      "votes": null
    },
    {
      "id": "956614",
      "postDate": "08/03/2020 16:19:54",
      "content": "<p>Seems like they are talking about histogram of isolated lung. See page 44, under \"Implementation Details\": \"First, the lungs were isolated from the mediastinum and chest wall using a commercially available, automated lung segmentation software\". Anybody found a good, open-source or pre-trained CNN for lung segmentation?</p>",
      "rawMarkdown": "Seems like they are talking about histogram of isolated lung. See page 44, under \"Implementation Details\": \"First, the lungs were isolated from the mediastinum and chest wall using a commercially available, automated lung segmentation software\". Anybody found a good, open-source or pre-trained CNN for lung segmentation?",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 949405,
      "author_name": "aadhavvignesh",
      "author_url": "",
      "post_date": "07/28/2020 16:11:01",
      "content": "<p>Thanks for posting the thesis! This definitely seems worth taking a look.</p>\n\n<p>I hope someone comes up with a real use-case of the histogram's kurtosis and skewness in their models.</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 950108,
      "author_name": "alexj21",
      "author_url": "",
      "post_date": "07/29/2020 07:28:19",
      "content": "<p>It was also tackled <a href=\"https://www.kaggle.com/c/osic-pulmonary-fibrosis-progression/discussion/165727\">here</a>\nIt looks promising. It's on my todo list</p>",
      "votes": null,
      "replies": [
        {
          "id": 950193,
          "author_name": "jameschapman19",
          "author_url": "",
          "post_date": "07/29/2020 08:41:58",
          "content": "<p>Ah thanks I'd managed to skip over that section! Well at least it's backed up twice now!</p>",
          "votes": null,
          "replies": []
        }
      ]
    },
    {
      "id": 956614,
      "author_name": "lirenzhu",
      "author_url": "",
      "post_date": "08/03/2020 16:19:54",
      "content": "<p>Seems like they are talking about histogram of isolated lung. See page 44, under \"Implementation Details\": \"First, the lungs were isolated from the mediastinum and chest wall using a commercially available, automated lung segmentation software\". Anybody found a good, open-source or pre-trained CNN for lung segmentation?</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "949309": "\"AI-driven Detection, Characterization and Classification\nof Chronic Lung Diseases\"\n\nThe whole thesis is probably quite useful.\n\nIn particular I'm looking at an interesting idea on page 45 showing that there is a difference in the histograms of healthy and diseased lungs. Potentially a nice handcrafted feature. I'll see if I can come up with a notebook demonstrating it unless someone beats me to it.",
    "949405": "Thanks for posting the thesis! This definitely seems worth taking a look.\n\nI hope someone comes up with a real use-case of the histogram's kurtosis and skewness in their models.",
    "950108": "It was also tackled [here](https://www.kaggle.com/c/osic-pulmonary-fibrosis-progression/discussion/165727)\nIt looks promising. It's on my todo list",
    "950193": "Ah thanks I'd managed to skip over that section! Well at least it's backed up twice now!",
    "956614": "Seems like they are talking about histogram of isolated lung. See page 44, under \"Implementation Details\": \"First, the lungs were isolated from the mediastinum and chest wall using a commercially available, automated lung segmentation software\". Anybody found a good, open-source or pre-trained CNN for lung segmentation?"
  },
  "source": "meta"
}