{
  "id": 164925,
  "title": "What does it mean by multiple dcm images for a single patient?",
  "url": "/competitions/osic-pulmonary-fibrosis-progression/discussion/164925",
  "author_name": "Durvesh Malpure",
  "post_date": "2020-07-07T21:37:24.079000",
  "votes": 13,
  "comment_count": 7,
  "views": 0,
  "content": "<p>Does it mean that they were taken on continuous weeks or do all of them combined make a CT scan? I'm very new to DICOM files. Any help is appreciated!</p>",
  "messages": [
    {
      "id": 919446,
      "postDate": "2020-07-07T21:37:24.080Z",
      "content": "<p>Does it mean that they were taken on continuous weeks or do all of them combined make a CT scan? I'm very new to DICOM files. Any help is appreciated!</p>",
      "rawMarkdown": "Does it mean that they were taken on continuous weeks or do all of them combined make a CT scan? I'm very new to DICOM files. Any help is appreciated!",
      "votes": 13
    },
    {
      "id": 919451,
      "postDate": "2020-07-07T21:43:24.777Z",
      "content": "<p>Multiple dcm images represent slices. CT imaging produces a 3D volume for each scan, this volume consists of 2D slices, each slice is a dcm image in our case. In other words, by stacking 2D images, you get the volume (case or patient). They're all taken at the same time. I hope this helps.</p>",
      "rawMarkdown": "Multiple dcm images represent slices. CT imaging produces a 3D volume for each scan, this volume consists of 2D slices, each slice is a dcm image in our case. In other words, by stacking 2D images, you get the volume (case or patient). They're all taken at the same time. I hope this helps.",
      "votes": 14,
      "replies": [
        {
          "id": 919458,
          "postDate": "2020-07-07T21:54:54.277Z",
          "content": "<p>Thanks a lot! It cleared my doubt.👌 </p>",
          "rawMarkdown": "Thanks a lot! It cleared my doubt.👌 "
        },
        {
          "id": 920580,
          "postDate": "2020-07-08T17:27:49.950Z",
          "content": "<p>I noticed that the number of slices for a given patient varies very widely (the range is from 12 to 1018, with a median of 98). Why is this? Are some image sets just denser along the z-axis? If so, is there a way we can know the exact z-coordinate of a given image?</p>",
          "rawMarkdown": "I noticed that the number of slices for a given patient varies very widely (the range is from 12 to 1018, with a median of 98). Why is this? Are some image sets just denser along the z-axis? If so, is there a way we can know the exact z-coordinate of a given image?",
          "votes": 2
        },
        {
          "id": 920587,
          "postDate": "2020-07-08T17:37:18.747Z",
          "content": "<p>Looking at a few cases, slice thickness ranges from sub-millimeter to 5 millimeters. This is typical in clinical practice. Depending on the reason for a study and local practice, you might store the sub-millimeter images. In other cases, you would just store the 5 millimeter slices.</p>\n\n<p>Also, a full study would often have both thin and thick slices and also Sagittal and Coronal reconstructions. I think this data only has axial images.</p>\n\n<p>Some of the studies with very few images could be either a study taken for followup where they only got a few images to lower radiation dose. I also suspect we might be purposely provided only some images to see if our algorithms can handle them.</p>\n\n<p>For the Z-axis, the Dicom headers have a position. You can take the change in Z-axis from image to image to calculate the inter-slice distance.</p>\n\n<p>For the few studies I looked at, the images appeared to be in order (if you account for the numeric sort order of 1.dcm, 10.dcm, 11.dcm ... 2.dcm (you have to resort taking numbers into account)).</p>\n\n<p>I think another thread suggested there were some studies that were out of order. You could re-order them based on the Z-axis data.</p>",
          "rawMarkdown": "Looking at a few cases, slice thickness ranges from sub-millimeter to 5 millimeters. This is typical in clinical practice. Depending on the reason for a study and local practice, you might store the sub-millimeter images. In other cases, you would just store the 5 millimeter slices.\n\nAlso, a full study would often have both thin and thick slices and also Sagittal and Coronal reconstructions. I think this data only has axial images.\n\nSome of the studies with very few images could be either a study taken for followup where they only got a few images to lower radiation dose. I also suspect we might be purposely provided only some images to see if our algorithms can handle them.\n\nFor the Z-axis, the Dicom headers have a position. You can take the change in Z-axis from image to image to calculate the inter-slice distance.\n\nFor the few studies I looked at, the images appeared to be in order (if you account for the numeric sort order of 1.dcm, 10.dcm, 11.dcm ... 2.dcm (you have to resort taking numbers into account)).\n\nI think another thread suggested there were some studies that were out of order. You could re-order them based on the Z-axis data.",
          "votes": 6
        },
        {
          "id": 920736,
          "postDate": "2020-07-08T19:14:53.367Z",
          "content": "<p>Just to clarify, would the Z-axis value be the value thus obtained?</p>\n\n<blockquote>\n  <p>f = pydicom.dcmread(filepath + filename)\n  f['ImagePositionPatient']._value[2]</p>\n</blockquote>\n\n<p>I'm not familiar with Pydicom Dataset objects but from my digging this looks like the values that would allow one to calculate the distance between slices (the unit of measurement is unclear but I don't think it really matters here).</p>",
          "rawMarkdown": "Just to clarify, would the Z-axis value be the value thus obtained?\n\n&gt; f = pydicom.dcmread(filepath + filename)\nf['ImagePositionPatient']._value[2]\n\nI'm not familiar with Pydicom Dataset objects but from my digging this looks like the values that would allow one to calculate the distance between slices (the unit of measurement is unclear but I don't think it really matters here)."
        },
        {
          "id": 920811,
          "postDate": "2020-07-08T20:06:23.193Z",
          "content": "<p>You are correct. The unit is \"millimeters\".</p>\n\n<p>All DICOM data is labeled with \"tags\" consisting of two hexadecimal numbers.</p>\n\n<p>For instance file:\n/input/train/ID00009637202177434476278/1.dcm</p>\n\n<p>Has the tags (excuse the poor formatting):</p>\n\n<p>Dataset.file_meta -------------------------------\n(0002, 0000) File Meta Information Group Length  UL: 206\n(0002, 0001) File Meta Information Version       OB: b'\\x00\\x01'\n(0002, 0002) Media Storage SOP Class UID         UI: CT Image Storage\n(0002, 0003) Media Storage SOP Instance UID      UI: 1.2.276.0.7230010.3.1.4.0.37492.1591423150.182195\n(0002, 0010) Transfer Syntax UID                 UI: Explicit VR Little Endian\n(0002, 0012) Implementation Class UID            UI: 1.2.276.0.7230010.3.0.3.6.1\n(0002, 0013) Implementation Version Name         SH: 'OSIRIX_361'</p>\n\n<h2>(0002, 0016) Source Application Entity Title     AE: 'ANONYMOUS'</h2>\n\n<p>(0008, 0018) SOP Instance UID                    UI: 1.2.276.0.7230010.3.1.4.0.37492.1591423150.182195\n(0008, 0060) Modality                            CS: 'CT'\n(0010, 0010) Patient's Name                      PN: 'ID00009637202177434476278'\n(0010, 0020) Patient ID                          LO: 'ID00009637202177434476278'\n(0010, 0040) Patient's Sex                       CS: ''\n(0018, 0015) Body Part Examined                  CS: 'Chest'\n(0020, 000d) Study Instance UID                  UI: 2.25.156962683457839326089809785890930019885\n(0020, 000e) Series Instance UID                 UI: 1.3.6.1.4.1.19291.2.1.2.11622117719213522261311595026142\n(0020, 0010) Study ID                            SH: ''\n(0020, 0013) Instance Number                     IS: \"1\"\n(0020, 0032) Image Position (Patient)            DS: [-171.634766, -333.634766, -37.000000]\n(0020, 0037) Image Orientation (Patient)         DS: [1.000000, 0.000000, 0.000000, 0.000000, 1.000000, 0.000000]\n(0028, 0002) Samples per Pixel                   US: 1\n(0028, 0004) Photometric Interpretation          CS: 'MONOCHROME2'\n(0028, 0010) Rows                                US: 768\n(0028, 0011) Columns                             US: 768\n(0028, 0030) Pixel Spacing                       DS: [0.486979, 0.486979]\n(0028, 0100) Bits Allocated                      US: 16\n(0028, 0101) Bits Stored                         US: 16\n(0028, 0102) High Bit                            US: 15\n(0028, 0103) Pixel Representation                US: 0\n(0028, 1050) Window Center                       DS: \"-500.0\"\n(0028, 1051) Window Width                        DS: \"1500.0\"\n(0028, 1052) Rescale Intercept                   DS: \"-1024.0\"\n(0028, 1053) Rescale Slope                       DS: \"1.0\"\n(7fe0, 0010) Pixel Data                          OW: Array of 1179648 elements</p>\n\n<p>-Rich</p>",
          "rawMarkdown": "You are correct. The unit is \"millimeters\".\n\nAll DICOM data is labeled with \"tags\" consisting of two hexadecimal numbers.\n\nFor instance file:\n/input/train/ID00009637202177434476278/1.dcm\n\nHas the tags (excuse the poor formatting):\n\nDataset.file_meta -------------------------------\n(0002, 0000) File Meta Information Group Length  UL: 206\n(0002, 0001) File Meta Information Version       OB: b'\\x00\\x01'\n(0002, 0002) Media Storage SOP Class UID         UI: CT Image Storage\n(0002, 0003) Media Storage SOP Instance UID      UI: 1.2.276.0.7230010.3.1.4.0.37492.1591423150.182195\n(0002, 0010) Transfer Syntax UID                 UI: Explicit VR Little Endian\n(0002, 0012) Implementation Class UID            UI: 1.2.276.0.7230010.3.0.3.6.1\n(0002, 0013) Implementation Version Name         SH: 'OSIRIX_361'\n(0002, 0016) Source Application Entity Title     AE: 'ANONYMOUS'\n-------------------------------------------------\n(0008, 0018) SOP Instance UID                    UI: 1.2.276.0.7230010.3.1.4.0.37492.1591423150.182195\n(0008, 0060) Modality                            CS: 'CT'\n(0010, 0010) Patient's Name                      PN: 'ID00009637202177434476278'\n(0010, 0020) Patient ID                          LO: 'ID00009637202177434476278'\n(0010, 0040) Patient's Sex                       CS: ''\n(0018, 0015) Body Part Examined                  CS: 'Chest'\n(0020, 000d) Study Instance UID                  UI: 2.25.156962683457839326089809785890930019885\n(0020, 000e) Series Instance UID                 UI: 1.3.6.1.4.1.19291.2.1.2.11622117719213522261311595026142\n(0020, 0010) Study ID                            SH: ''\n(0020, 0013) Instance Number                     IS: \"1\"\n(0020, 0032) Image Position (Patient)            DS: [-171.634766, -333.634766, -37.000000]\n(0020, 0037) Image Orientation (Patient)         DS: [1.000000, 0.000000, 0.000000, 0.000000, 1.000000, 0.000000]\n(0028, 0002) Samples per Pixel                   US: 1\n(0028, 0004) Photometric Interpretation          CS: 'MONOCHROME2'\n(0028, 0010) Rows                                US: 768\n(0028, 0011) Columns                             US: 768\n(0028, 0030) Pixel Spacing                       DS: [0.486979, 0.486979]\n(0028, 0100) Bits Allocated                      US: 16\n(0028, 0101) Bits Stored                         US: 16\n(0028, 0102) High Bit                            US: 15\n(0028, 0103) Pixel Representation                US: 0\n(0028, 1050) Window Center                       DS: \"-500.0\"\n(0028, 1051) Window Width                        DS: \"1500.0\"\n(0028, 1052) Rescale Intercept                   DS: \"-1024.0\"\n(0028, 1053) Rescale Slope                       DS: \"1.0\"\n(7fe0, 0010) Pixel Data                          OW: Array of 1179648 elements\n\n-Rich",
          "votes": 1
        },
        {
          "id": 984852,
          "postDate": "2020-08-25T10:28:24.200Z",
          "rawMarkdown": "",
          "isDeleted": true
        }
      ]
    }
  ],
  "comments": [
    {
      "id": 919451,
      "author_name": "Ahmed Shahin",
      "author_url": "",
      "post_date": "2020-07-07T21:43:24.777000",
      "content": "<p>Multiple dcm images represent slices. CT imaging produces a 3D volume for each scan, this volume consists of 2D slices, each slice is a dcm image in our case. In other words, by stacking 2D images, you get the volume (case or patient). They're all taken at the same time. I hope this helps.</p>",
      "votes": 14,
      "replies": [
        {
          "id": 919458,
          "author_name": "Durvesh Malpure",
          "author_url": "",
          "post_date": "2020-07-07T21:54:54.277000",
          "content": "<p>Thanks a lot! It cleared my doubt.👌 </p>",
          "votes": 0,
          "replies": []
        },
        {
          "id": 920580,
          "author_name": "Mathieu Beaudoin",
          "author_url": "",
          "post_date": "2020-07-08T17:27:49.950000",
          "content": "<p>I noticed that the number of slices for a given patient varies very widely (the range is from 12 to 1018, with a median of 98). Why is this? Are some image sets just denser along the z-axis? If so, is there a way we can know the exact z-coordinate of a given image?</p>",
          "votes": 2,
          "replies": []
        },
        {
          "id": 920587,
          "author_name": "quadcore/Richard Epstein",
          "author_url": "",
          "post_date": "2020-07-08T17:37:18.747000",
          "content": "<p>Looking at a few cases, slice thickness ranges from sub-millimeter to 5 millimeters. This is typical in clinical practice. Depending on the reason for a study and local practice, you might store the sub-millimeter images. In other cases, you would just store the 5 millimeter slices.</p>\n\n<p>Also, a full study would often have both thin and thick slices and also Sagittal and Coronal reconstructions. I think this data only has axial images.</p>\n\n<p>Some of the studies with very few images could be either a study taken for followup where they only got a few images to lower radiation dose. I also suspect we might be purposely provided only some images to see if our algorithms can handle them.</p>\n\n<p>For the Z-axis, the Dicom headers have a position. You can take the change in Z-axis from image to image to calculate the inter-slice distance.</p>\n\n<p>For the few studies I looked at, the images appeared to be in order (if you account for the numeric sort order of 1.dcm, 10.dcm, 11.dcm ... 2.dcm (you have to resort taking numbers into account)).</p>\n\n<p>I think another thread suggested there were some studies that were out of order. You could re-order them based on the Z-axis data.</p>",
          "votes": 6,
          "replies": []
        },
        {
          "id": 920736,
          "author_name": "Mathieu Beaudoin",
          "author_url": "",
          "post_date": "2020-07-08T19:14:53.367000",
          "content": "<p>Just to clarify, would the Z-axis value be the value thus obtained?</p>\n\n<blockquote>\n  <p>f = pydicom.dcmread(filepath + filename)\n  f['ImagePositionPatient']._value[2]</p>\n</blockquote>\n\n<p>I'm not familiar with Pydicom Dataset objects but from my digging this looks like the values that would allow one to calculate the distance between slices (the unit of measurement is unclear but I don't think it really matters here).</p>",
          "votes": 0,
          "replies": []
        },
        {
          "id": 920811,
          "author_name": "quadcore/Richard Epstein",
          "author_url": "",
          "post_date": "2020-07-08T20:06:23.193000",
          "content": "<p>You are correct. The unit is \"millimeters\".</p>\n\n<p>All DICOM data is labeled with \"tags\" consisting of two hexadecimal numbers.</p>\n\n<p>For instance file:\n/input/train/ID00009637202177434476278/1.dcm</p>\n\n<p>Has the tags (excuse the poor formatting):</p>\n\n<p>Dataset.file_meta -------------------------------\n(0002, 0000) File Meta Information Group Length  UL: 206\n(0002, 0001) File Meta Information Version       OB: b'\\x00\\x01'\n(0002, 0002) Media Storage SOP Class UID         UI: CT Image Storage\n(0002, 0003) Media Storage SOP Instance UID      UI: 1.2.276.0.7230010.3.1.4.0.37492.1591423150.182195\n(0002, 0010) Transfer Syntax UID                 UI: Explicit VR Little Endian\n(0002, 0012) Implementation Class UID            UI: 1.2.276.0.7230010.3.0.3.6.1\n(0002, 0013) Implementation Version Name         SH: 'OSIRIX_361'</p>\n\n<h2>(0002, 0016) Source Application Entity Title     AE: 'ANONYMOUS'</h2>\n\n<p>(0008, 0018) SOP Instance UID                    UI: 1.2.276.0.7230010.3.1.4.0.37492.1591423150.182195\n(0008, 0060) Modality                            CS: 'CT'\n(0010, 0010) Patient's Name                      PN: 'ID00009637202177434476278'\n(0010, 0020) Patient ID                          LO: 'ID00009637202177434476278'\n(0010, 0040) Patient's Sex                       CS: ''\n(0018, 0015) Body Part Examined                  CS: 'Chest'\n(0020, 000d) Study Instance UID                  UI: 2.25.156962683457839326089809785890930019885\n(0020, 000e) Series Instance UID                 UI: 1.3.6.1.4.1.19291.2.1.2.11622117719213522261311595026142\n(0020, 0010) Study ID                            SH: ''\n(0020, 0013) Instance Number                     IS: \"1\"\n(0020, 0032) Image Position (Patient)            DS: [-171.634766, -333.634766, -37.000000]\n(0020, 0037) Image Orientation (Patient)         DS: [1.000000, 0.000000, 0.000000, 0.000000, 1.000000, 0.000000]\n(0028, 0002) Samples per Pixel                   US: 1\n(0028, 0004) Photometric Interpretation          CS: 'MONOCHROME2'\n(0028, 0010) Rows                                US: 768\n(0028, 0011) Columns                             US: 768\n(0028, 0030) Pixel Spacing                       DS: [0.486979, 0.486979]\n(0028, 0100) Bits Allocated                      US: 16\n(0028, 0101) Bits Stored                         US: 16\n(0028, 0102) High Bit                            US: 15\n(0028, 0103) Pixel Representation                US: 0\n(0028, 1050) Window Center                       DS: \"-500.0\"\n(0028, 1051) Window Width                        DS: \"1500.0\"\n(0028, 1052) Rescale Intercept                   DS: \"-1024.0\"\n(0028, 1053) Rescale Slope                       DS: \"1.0\"\n(7fe0, 0010) Pixel Data                          OW: Array of 1179648 elements</p>\n\n<p>-Rich</p>",
          "votes": 1,
          "replies": []
        },
        {
          "id": 984852,
          "author_name": "",
          "author_url": "",
          "post_date": "2020-08-25T10:28:24.200000",
          "content": "",
          "votes": 0,
          "replies": []
        }
      ]
    }
  ],
  "raw_markdown_by_id": {
    "919446": "Does it mean that they were taken on continuous weeks or do all of them combined make a CT scan? I'm very new to DICOM files. Any help is appreciated!",
    "919451": "Multiple dcm images represent slices. CT imaging produces a 3D volume for each scan, this volume consists of 2D slices, each slice is a dcm image in our case. In other words, by stacking 2D images, you get the volume (case or patient). They're all taken at the same time. I hope this helps."
  }
}