{ "id": 210064, "title": "wrong annotation", "url": "/competitions/ranzcr-clip-catheter-line-classification/discussion/210064", "author_name": "hengck23", "post_date": "2021-01-09T14:43:44.396000", "votes": 38, "comment_count": 13, "views": 0, "content": "

these are wrong annotation discovered by my algorithm

\n

\"\"

\n

1.2.826.0.1.3680043.8.498.93345761486297843389996628528592497280.png

", "messages": [ { "id": 1146102, "postDate": "2021-01-09T14:43:44.397Z", "content": "

these are wrong annotation discovered by my algorithm

\n

\"\"

\n

1.2.826.0.1.3680043.8.498.93345761486297843389996628528592497280.png

", "rawMarkdown": "these are wrong annotation discovered by my algorithm\n\n![](https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F113660%2F6d3ce67a13eb7cd85b901d9a56757990%2F1.2.826.0.1.3680043.8.498.93345761486297843389996628528592497280.png?generation=1610203300256854&alt=media)\n\n1.2.826.0.1.3680043.8.498.93345761486297843389996628528592497280.png", "votes": 38 }, { "id": 1146111, "postDate": "2021-01-09T14:48:33.947Z", "content": "

This is my collection so far (index represents row index of train_annotations.csv):

\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n
indexStudyInstanceUIDlabelnew_label
035891.2.826.0.1.3680043.8.498.57005638787237813934531972491254580369CVC - BorderlineNGT - Borderline
143441.2.826.0.1.3680043.8.498.93345761486297843389996628528592497280ETT - AbnormalCVC - Abnormal
262941.2.826.0.1.3680043.8.498.50891603479257167332052859560303996365NGT - NormalCVC - Normal
375581.2.826.0.1.3680043.8.498.32665013930528750130301395098139968929NGT - BorderlineCVC - Borderline
484571.2.826.0.1.3680043.8.498.47822809495672253227315400926882161159NGT - BorderlineCVC - Borderline
585861.2.826.0.1.3680043.8.498.55171965195784371324650309161724846475NGT - BorderlineCVC - Borderline
685891.2.826.0.1.3680043.8.498.29639870594803047496855371142714987539ETT - NormalCVC - Normal
799081.2.826.0.1.3680043.8.498.52422864792637441690285442425747003963NGT - NormalETT - Normal
8108891.2.826.0.1.3680043.8.498.51277351337858188519077141427236143108NGT - NormalCVC - Normal
9109631.2.826.0.1.3680043.8.498.33011244702337270174558484639492100815CVC - NormalNGT - Normal
10119021.2.826.0.1.3680043.8.498.10505287747515183956922280117689383476NGT - NormalCVC - Normal
11120411.2.826.0.1.3680043.8.498.43340424479611237895060478106689360500NGT - NormalCVC - Normal
12127821.2.826.0.1.3680043.8.498.12545979153892772426852721449004507757NGT - AbnormalCVC - Abnormal
13135131.2.826.0.1.3680043.8.498.83700037297895094021306651705503600111NGT - NormalETT - Normal
14142261.2.826.0.1.3680043.8.498.35772244095675958072394978496245125294NGT - NormalETT - Normal
15157501.2.826.0.1.3680043.8.498.96130195933728659348647733812659169362CVC - AbnormalNGT - Abnormal
16157791.2.826.0.1.3680043.8.498.75269816256944932004789976844599885553NGT - AbnormalCVC - Abnormal
17166291.2.826.0.1.3680043.8.498.11935284122896798228836385959451625327NGT - AbnormalCVC - Abnormal
18175011.2.826.0.1.3680043.8.498.83574817573978660270935463700320068005NGT - AbnormalCVC - Abnormal
", "rawMarkdown": "This is my collection so far (`index` represents row index of `train_annotations.csv`):\n\n| | index | StudyInstanceUID | label | new_label |\n|---:|--------:|:-----------------------------------------------------------------|:-----------------|:-----------------|\n| 0 | 3589 | 1.2.826.0.1.3680043.8.498.57005638787237813934531972491254580369 | CVC - Borderline | NGT - Borderline |\n| 1 | 4344 | 1.2.826.0.1.3680043.8.498.93345761486297843389996628528592497280 | ETT - Abnormal | CVC - Abnormal |\n| 2 | 6294 | 1.2.826.0.1.3680043.8.498.50891603479257167332052859560303996365 | NGT - Normal | CVC - Normal |\n| 3 | 7558 | 1.2.826.0.1.3680043.8.498.32665013930528750130301395098139968929 | NGT - Borderline | CVC - Borderline |\n| 4 | 8457 | 1.2.826.0.1.3680043.8.498.47822809495672253227315400926882161159 | NGT - Borderline | CVC - Borderline |\n| 5 | 8586 | 1.2.826.0.1.3680043.8.498.55171965195784371324650309161724846475 | NGT - Borderline | CVC - Borderline |\n| 6 | 8589 | 1.2.826.0.1.3680043.8.498.29639870594803047496855371142714987539 | ETT - Normal | CVC - Normal |\n| 7 | 9908 | 1.2.826.0.1.3680043.8.498.52422864792637441690285442425747003963 | NGT - Normal | ETT - Normal |\n| 8 | 10889 | 1.2.826.0.1.3680043.8.498.51277351337858188519077141427236143108 | NGT - Normal | CVC - Normal |\n| 9 | 10963 | 1.2.826.0.1.3680043.8.498.33011244702337270174558484639492100815 | CVC - Normal | NGT - Normal |\n| 10 | 11902 | 1.2.826.0.1.3680043.8.498.10505287747515183956922280117689383476 | NGT - Normal | CVC - Normal |\n| 11 | 12041 | 1.2.826.0.1.3680043.8.498.43340424479611237895060478106689360500 | NGT - Normal | CVC - Normal |\n| 12 | 12782 | 1.2.826.0.1.3680043.8.498.12545979153892772426852721449004507757 | NGT - Abnormal | CVC - Abnormal |\n| 13 | 13513 | 1.2.826.0.1.3680043.8.498.83700037297895094021306651705503600111 | NGT - Normal | ETT - Normal |\n| 14 | 14226 | 1.2.826.0.1.3680043.8.498.35772244095675958072394978496245125294 | NGT - Normal | ETT - Normal |\n| 15 | 15750 | 1.2.826.0.1.3680043.8.498.96130195933728659348647733812659169362 | CVC - Abnormal | NGT - Abnormal |\n| 16 | 15779 | 1.2.826.0.1.3680043.8.498.75269816256944932004789976844599885553 | NGT - Abnormal | CVC - Abnormal |\n| 17 | 16629 | 1.2.826.0.1.3680043.8.498.11935284122896798228836385959451625327 | NGT - Abnormal | CVC - Abnormal |\n| 18 | 17501 | 1.2.826.0.1.3680043.8.498.83574817573978660270935463700320068005 | NGT - Abnormal | CVC - Abnormal |\n", "votes": 31, "replies": [ { "id": 1188437, "postDate": "2021-02-06T08:33:56.733Z", "rawMarkdown": "", "isDeleted": true } ] }, { "id": 1146115, "postDate": "2021-01-09T14:51:23.893Z", "content": "

Also, inverted images:

\n

bad_imgs = ['1.2.826.0.1.3680043.8.498.45667033584171921001890022815707001978',
\n '1.2.826.0.1.3680043.8.498.93053605340693492468203536922883055634',
\n '1.2.826.0.1.3680043.8.498.46805490620878014733000033680286522306',
\n '1.2.826.0.1.3680043.8.498.60784708544708592859086705347710043758',
\n ]

", "rawMarkdown": "Also, inverted images:\n\nbad_imgs = ['1.2.826.0.1.3680043.8.498.45667033584171921001890022815707001978',\n '1.2.826.0.1.3680043.8.498.93053605340693492468203536922883055634',\n '1.2.826.0.1.3680043.8.498.46805490620878014733000033680286522306',\n '1.2.826.0.1.3680043.8.498.60784708544708592859086705347710043758',\n ]", "votes": 9, "replies": [ { "id": 1149491, "postDate": "2021-01-11T21:52:57.357Z", "content": "

My extended list of inverted images:

\n

['1.2.826.0.1.3680043.8.498.93053605340693492468203536922883055634',
\n'1.2.826.0.1.3680043.8.498.89369371707034087254309411362762932453',
\n'1.2.826.0.1.3680043.8.498.40874263518848015471042617691509326469',
\n'1.2.826.0.1.3680043.8.498.60784708544708592859086705347710043758',
\n'1.2.826.0.1.3680043.8.498.85400081110981214468425786540292202327',
\n'1.2.826.0.1.3680043.8.498.46805490620878014733000033680286522306',
\n'1.2.826.0.1.3680043.8.498.12426606037326639593164801231383702795',
\n'1.2.826.0.1.3680043.8.498.44981676356222715641792310487558318854',
\n'1.2.826.0.1.3680043.8.498.92223955132523718945948760352349399544',
\n'1.2.826.0.1.3680043.8.498.45667033584171921001890022815707001978',
\n'1.2.826.0.1.3680043.8.498.62409687695240227890004324406035594441']

", "rawMarkdown": "My extended list of inverted images:\n\n['1.2.826.0.1.3680043.8.498.93053605340693492468203536922883055634',\n'1.2.826.0.1.3680043.8.498.89369371707034087254309411362762932453',\n'1.2.826.0.1.3680043.8.498.40874263518848015471042617691509326469',\n'1.2.826.0.1.3680043.8.498.60784708544708592859086705347710043758',\n'1.2.826.0.1.3680043.8.498.85400081110981214468425786540292202327',\n'1.2.826.0.1.3680043.8.498.46805490620878014733000033680286522306',\n'1.2.826.0.1.3680043.8.498.12426606037326639593164801231383702795',\n'1.2.826.0.1.3680043.8.498.44981676356222715641792310487558318854',\n'1.2.826.0.1.3680043.8.498.92223955132523718945948760352349399544',\n'1.2.826.0.1.3680043.8.498.45667033584171921001890022815707001978',\n'1.2.826.0.1.3680043.8.498.62409687695240227890004324406035594441']", "votes": 8 }, { "id": 1219961, "postDate": "2021-02-27T11:56:16.857Z", "content": "

Nice list. I tried to find out inverted images with this notebook and no other image found than yours. Accidentally, I found some bad-oriented images like upside-down.
\n[
\n \"1.2.826.0.1.3680043.8.498.56964749951381900643748134536978560792.jpg\",
\n \"1.2.826.0.1.3680043.8.498.76129162274163220380041920805275862370.jpg\",
\n \"1.2.826.0.1.3680043.8.498.11582767971938057384592968535311883741.jpg\",
\n \"1.2.826.0.1.3680043.8.498.29092351703254040179552572484602410700.jpg\",
\n \"1.2.826.0.1.3680043.8.498.11620053814932996350746126485322079242.jpg\",
\n]

", "rawMarkdown": "Nice list. I tried to find out inverted images with [this notebook](https://www.kaggle.com/york33/find-out-inverted-images) and no other image found than yours. Accidentally, I found some bad-oriented images like upside-down.\n[\n \"1.2.826.0.1.3680043.8.498.56964749951381900643748134536978560792.jpg\",\n \"1.2.826.0.1.3680043.8.498.76129162274163220380041920805275862370.jpg\",\n \"1.2.826.0.1.3680043.8.498.11582767971938057384592968535311883741.jpg\",\n \"1.2.826.0.1.3680043.8.498.29092351703254040179552572484602410700.jpg\",\n \"1.2.826.0.1.3680043.8.498.11620053814932996350746126485322079242.jpg\",\n]", "votes": 1 }, { "id": 1220465, "postDate": "2021-02-28T02:58:38.780Z", "content": "

Hello,how to process the inverted images?

", "rawMarkdown": "Hello,how to process the inverted images?" }, { "id": 1220482, "postDate": "2021-02-28T03:57:39.427Z", "content": "

To training data, just re-invert the known inverted images. But test data at submission time may contain inverted images. In that case, you need to detect if it is inverted with some ML model or some other statistic logic like using the color histogram. My model is still weak so I don't use that.
\nBut it might be not a big problem. It depends on how many inverted(or bad oriented)images in test data at submission time.
\nOr, as an augmentation, you can intentionally invert some of the training image to make the model immune to inverted images.

", "rawMarkdown": "To training data, just re-invert the known inverted images. But test data at submission time may contain inverted images. In that case, you need to detect if it is inverted with some ML model or some other statistic logic like using the color histogram. [My model](https://www.kaggle.com/york33/find-out-inverted-images) is still weak so I don't use that.\nBut it might be not a big problem. It depends on how many inverted(or bad oriented)images in test data at submission time.\nOr, as an augmentation, you can intentionally invert some of the training image to make the model immune to inverted images." } ] }, { "id": 1146105, "postDate": "2021-01-09T14:46:18.783Z", "content": "

1.2.826.0.1.3680043.8.498.57005638787237813934531972491254580369[1].png

\n

this is annotated as cvc-borderline (which i think is wrong)

\n

\"\"

", "rawMarkdown": "1.2.826.0.1.3680043.8.498.57005638787237813934531972491254580369[1].png\n\nthis is annotated as cvc-borderline (which i think is wrong)\n\n![](https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F113660%2F2f9b474acdfd68f3e67cc54c428ef407%2F1.2.826.0.1.3680043.8.498.570056387872378139345319724912545803691.png?generation=1610203555248556&alt=media)\n", "votes": 5, "replies": [ { "id": 1179037, "postDate": "2021-01-31T09:16:41.513Z", "content": "

would u mind sharing ur algorithm for wrong annotation detection?

", "rawMarkdown": "would u mind sharing ur algorithm for wrong annotation detection?" }, { "id": 1220462, "postDate": "2021-02-28T02:54:14.907Z", "content": "

Hi,how to detect them?

", "rawMarkdown": "Hi,how to detect them?" } ] }, { "id": 1168790, "postDate": "2021-01-25T07:14:35.820Z", "content": "

@raddar @hypocrites how to work with the inverted images？abandon it？

", "rawMarkdown": "@raddar @hypocrites how to work with the inverted images？abandon it？", "replies": [ { "id": 1168928, "postDate": "2021-01-25T08:54:04.980Z", "content": "

Hard to say.. I leave them as it is

", "rawMarkdown": "Hard to say.. I leave them as it is", "votes": 2 } ] }, { "id": 1148425, "postDate": "2021-01-11T05:56:39.533Z", "content": "

Hey @hengck23, can you give some intuition about this algorithm?

", "rawMarkdown": "Hey @hengck23, can you give some intuition about this algorithm?" } ], "comments": [ { "id": 1146111, "author_name": "raddar", "author_url": "", "post_date": "2021-01-09T14:48:33.947000", "content": "

This is my collection so far (index represents row index of train_annotations.csv):

\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n
indexStudyInstanceUIDlabelnew_label
035891.2.826.0.1.3680043.8.498.57005638787237813934531972491254580369CVC - BorderlineNGT - Borderline
143441.2.826.0.1.3680043.8.498.93345761486297843389996628528592497280ETT - AbnormalCVC - Abnormal
262941.2.826.0.1.3680043.8.498.50891603479257167332052859560303996365NGT - NormalCVC - Normal
375581.2.826.0.1.3680043.8.498.32665013930528750130301395098139968929NGT - BorderlineCVC - Borderline
484571.2.826.0.1.3680043.8.498.47822809495672253227315400926882161159NGT - BorderlineCVC - Borderline
585861.2.826.0.1.3680043.8.498.55171965195784371324650309161724846475NGT - BorderlineCVC - Borderline
685891.2.826.0.1.3680043.8.498.29639870594803047496855371142714987539ETT - NormalCVC - Normal
799081.2.826.0.1.3680043.8.498.52422864792637441690285442425747003963NGT - NormalETT - Normal
8108891.2.826.0.1.3680043.8.498.51277351337858188519077141427236143108NGT - NormalCVC - Normal
9109631.2.826.0.1.3680043.8.498.33011244702337270174558484639492100815CVC - NormalNGT - Normal
10119021.2.826.0.1.3680043.8.498.10505287747515183956922280117689383476NGT - NormalCVC - Normal
11120411.2.826.0.1.3680043.8.498.43340424479611237895060478106689360500NGT - NormalCVC - Normal
12127821.2.826.0.1.3680043.8.498.12545979153892772426852721449004507757NGT - AbnormalCVC - Abnormal
13135131.2.826.0.1.3680043.8.498.83700037297895094021306651705503600111NGT - NormalETT - Normal
14142261.2.826.0.1.3680043.8.498.35772244095675958072394978496245125294NGT - NormalETT - Normal
15157501.2.826.0.1.3680043.8.498.96130195933728659348647733812659169362CVC - AbnormalNGT - Abnormal
16157791.2.826.0.1.3680043.8.498.75269816256944932004789976844599885553NGT - AbnormalCVC - Abnormal
17166291.2.826.0.1.3680043.8.498.11935284122896798228836385959451625327NGT - AbnormalCVC - Abnormal
18175011.2.826.0.1.3680043.8.498.83574817573978660270935463700320068005NGT - AbnormalCVC - Abnormal
", "votes": 31, "replies": [ { "id": 1188437, "author_name": "", "author_url": "", "post_date": "2021-02-06T08:33:56.733000", "content": "", "votes": 0, "replies": [] } ] }, { "id": 1146115, "author_name": "raddar", "author_url": "", "post_date": "2021-01-09T14:51:23.893000", "content": "

Also, inverted images:

\n

bad_imgs = ['1.2.826.0.1.3680043.8.498.45667033584171921001890022815707001978',
\n '1.2.826.0.1.3680043.8.498.93053605340693492468203536922883055634',
\n '1.2.826.0.1.3680043.8.498.46805490620878014733000033680286522306',
\n '1.2.826.0.1.3680043.8.498.60784708544708592859086705347710043758',
\n ]

", "votes": 9, "replies": [ { "id": 1149491, "author_name": "anthony", "author_url": "", "post_date": "2021-01-11T21:52:57.357000", "content": "

My extended list of inverted images:

\n

['1.2.826.0.1.3680043.8.498.93053605340693492468203536922883055634',
\n'1.2.826.0.1.3680043.8.498.89369371707034087254309411362762932453',
\n'1.2.826.0.1.3680043.8.498.40874263518848015471042617691509326469',
\n'1.2.826.0.1.3680043.8.498.60784708544708592859086705347710043758',
\n'1.2.826.0.1.3680043.8.498.85400081110981214468425786540292202327',
\n'1.2.826.0.1.3680043.8.498.46805490620878014733000033680286522306',
\n'1.2.826.0.1.3680043.8.498.12426606037326639593164801231383702795',
\n'1.2.826.0.1.3680043.8.498.44981676356222715641792310487558318854',
\n'1.2.826.0.1.3680043.8.498.92223955132523718945948760352349399544',
\n'1.2.826.0.1.3680043.8.498.45667033584171921001890022815707001978',
\n'1.2.826.0.1.3680043.8.498.62409687695240227890004324406035594441']

", "votes": 8, "replies": [] }, { "id": 1219961, "author_name": "york33", "author_url": "", "post_date": "2021-02-27T11:56:16.857000", "content": "

Nice list. I tried to find out inverted images with this notebook and no other image found than yours. Accidentally, I found some bad-oriented images like upside-down.
\n[
\n \"1.2.826.0.1.3680043.8.498.56964749951381900643748134536978560792.jpg\",
\n \"1.2.826.0.1.3680043.8.498.76129162274163220380041920805275862370.jpg\",
\n \"1.2.826.0.1.3680043.8.498.11582767971938057384592968535311883741.jpg\",
\n \"1.2.826.0.1.3680043.8.498.29092351703254040179552572484602410700.jpg\",
\n \"1.2.826.0.1.3680043.8.498.11620053814932996350746126485322079242.jpg\",
\n]

", "votes": 1, "replies": [] }, { "id": 1220465, "author_name": "Zekun", "author_url": "", "post_date": "2021-02-28T02:58:38.780000", "content": "

Hello,how to process the inverted images?

", "votes": 0, "replies": [] }, { "id": 1220482, "author_name": "york33", "author_url": "", "post_date": "2021-02-28T03:57:39.427000", "content": "

To training data, just re-invert the known inverted images. But test data at submission time may contain inverted images. In that case, you need to detect if it is inverted with some ML model or some other statistic logic like using the color histogram. My model is still weak so I don't use that.
\nBut it might be not a big problem. It depends on how many inverted(or bad oriented)images in test data at submission time.
\nOr, as an augmentation, you can intentionally invert some of the training image to make the model immune to inverted images.

", "votes": 0, "replies": [] } ] }, { "id": 1146105, "author_name": "hengck23", "author_url": "", "post_date": "2021-01-09T14:46:18.783000", "content": "

1.2.826.0.1.3680043.8.498.57005638787237813934531972491254580369[1].png

\n

this is annotated as cvc-borderline (which i think is wrong)

\n

\"\"

", "votes": 5, "replies": [ { "id": 1179037, "author_name": "Mohammed Sabry", "author_url": "", "post_date": "2021-01-31T09:16:41.513000", "content": "

would u mind sharing ur algorithm for wrong annotation detection?

", "votes": 0, "replies": [] }, { "id": 1220462, "author_name": "Zekun", "author_url": "", "post_date": "2021-02-28T02:54:14.907000", "content": "

Hi,how to detect them?

", "votes": 0, "replies": [] } ] }, { "id": 1168790, "author_name": "ZZemily", "author_url": "", "post_date": "2021-01-25T07:14:35.820000", "content": "

@raddar @hypocrites how to work with the inverted images？abandon it？

", "votes": 0, "replies": [ { "id": 1168928, "author_name": "raddar", "author_url": "", "post_date": "2021-01-25T08:54:04.980000", "content": "

Hard to say.. I leave them as it is

", "votes": 2, "replies": [] } ] }, { "id": 1148425, "author_name": "Akhilesh D. Kapse", "author_url": "", "post_date": "2021-01-11T05:56:39.533000", "content": "

Hey @hengck23, can you give some intuition about this algorithm?

", "votes": 0, "replies": [] } ], "raw_markdown_by_id": { "1146102": "these are wrong annotation discovered by my algorithm\n\n![](https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F113660%2F6d3ce67a13eb7cd85b901d9a56757990%2F1.2.826.0.1.3680043.8.498.93345761486297843389996628528592497280.png?generation=1610203300256854&alt=media)\n\n1.2.826.0.1.3680043.8.498.93345761486297843389996628528592497280.png", "1146111": "This is my collection so far (`index` represents row index of `train_annotations.csv`):\n\n| | index | StudyInstanceUID | label | new_label |\n|---:|--------:|:-----------------------------------------------------------------|:-----------------|:-----------------|\n| 0 | 3589 | 1.2.826.0.1.3680043.8.498.57005638787237813934531972491254580369 | CVC - Borderline | NGT - Borderline |\n| 1 | 4344 | 1.2.826.0.1.3680043.8.498.93345761486297843389996628528592497280 | ETT - Abnormal | CVC - Abnormal |\n| 2 | 6294 | 1.2.826.0.1.3680043.8.498.50891603479257167332052859560303996365 | NGT - Normal | CVC - Normal |\n| 3 | 7558 | 1.2.826.0.1.3680043.8.498.32665013930528750130301395098139968929 | NGT - Borderline | CVC - Borderline |\n| 4 | 8457 | 1.2.826.0.1.3680043.8.498.47822809495672253227315400926882161159 | NGT - Borderline | CVC - Borderline |\n| 5 | 8586 | 1.2.826.0.1.3680043.8.498.55171965195784371324650309161724846475 | NGT - Borderline | CVC - Borderline |\n| 6 | 8589 | 1.2.826.0.1.3680043.8.498.29639870594803047496855371142714987539 | ETT - Normal | CVC - Normal |\n| 7 | 9908 | 1.2.826.0.1.3680043.8.498.52422864792637441690285442425747003963 | NGT - Normal | ETT - Normal |\n| 8 | 10889 | 1.2.826.0.1.3680043.8.498.51277351337858188519077141427236143108 | NGT - Normal | CVC - Normal |\n| 9 | 10963 | 1.2.826.0.1.3680043.8.498.33011244702337270174558484639492100815 | CVC - Normal | NGT - Normal |\n| 10 | 11902 | 1.2.826.0.1.3680043.8.498.10505287747515183956922280117689383476 | NGT - Normal | CVC - Normal |\n| 11 | 12041 | 1.2.826.0.1.3680043.8.498.43340424479611237895060478106689360500 | NGT - Normal | CVC - Normal |\n| 12 | 12782 | 1.2.826.0.1.3680043.8.498.12545979153892772426852721449004507757 | NGT - Abnormal | CVC - Abnormal |\n| 13 | 13513 | 1.2.826.0.1.3680043.8.498.83700037297895094021306651705503600111 | NGT - Normal | ETT - Normal |\n| 14 | 14226 | 1.2.826.0.1.3680043.8.498.35772244095675958072394978496245125294 | NGT - Normal | ETT - Normal |\n| 15 | 15750 | 1.2.826.0.1.3680043.8.498.96130195933728659348647733812659169362 | CVC - Abnormal | NGT - Abnormal |\n| 16 | 15779 | 1.2.826.0.1.3680043.8.498.75269816256944932004789976844599885553 | NGT - Abnormal | CVC - Abnormal |\n| 17 | 16629 | 1.2.826.0.1.3680043.8.498.11935284122896798228836385959451625327 | NGT - Abnormal | CVC - Abnormal |\n| 18 | 17501 | 1.2.826.0.1.3680043.8.498.83574817573978660270935463700320068005 | NGT - Abnormal | CVC - Abnormal |\n", "1146115": "Also, inverted images:\n\nbad_imgs = ['1.2.826.0.1.3680043.8.498.45667033584171921001890022815707001978',\n '1.2.826.0.1.3680043.8.498.93053605340693492468203536922883055634',\n '1.2.826.0.1.3680043.8.498.46805490620878014733000033680286522306',\n '1.2.826.0.1.3680043.8.498.60784708544708592859086705347710043758',\n ]", "1146105": "1.2.826.0.1.3680043.8.498.57005638787237813934531972491254580369[1].png\n\nthis is annotated as cvc-borderline (which i think is wrong)\n\n![](https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F113660%2F2f9b474acdfd68f3e67cc54c428ef407%2F1.2.826.0.1.3680043.8.498.570056387872378139345319724912545803691.png?generation=1610203555248556&alt=media)\n", "1168790": "@raddar @hypocrites how to work with the inverted images？abandon it？", "1148425": "Hey @hengck23, can you give some intuition about this algorithm?" } }