![\"\"](\"https://www.googleapis.com/download/storage/v1/b/kaggle-user-content/o/inbox%2F113660%2Fdcb643c727d9fa03e26acd9a4d6fdfb2%2FSelection_054.png?generation=1591041058574098&alt=media\")
This dataset has many records for the same patient having varied age values. For example, Patient ID IP_0038545 has 28 records, having age values varying from 45 to 60. Shall we take it as erroneous data or the records themselves taken at different ages ?
", "messages": [ { "id": "870369", "postDate": "06/01/2020 16:43:37", "content": "This dataset has many records for the same patient having varied age values. For example, Patient ID IP_0038545 has 28 records, having age values varying from 45 to 60. Shall we take it as erroneous data or the records themselves taken at different ages ?
", "rawMarkdown": "This dataset has many records for the same patient having varied age values. For example, Patient ID IP_0038545 has 28 records, having age values varying from 45 to 60. Shall we take it as erroneous data or the records themselves taken at different ages ?", "votes": null }, { "id": "870406", "postDate": "06/01/2020 16:58:39", "content": "This interests me
", "rawMarkdown": "This interests me", "votes": null }, { "id": "870417", "postDate": "06/01/2020 17:04:00", "content": "Is it possible that he has been tested for several years? Or does he have very different ages?
", "rawMarkdown": "Is it possible that he has been tested for several years? Or does he have very different ages?", "votes": null }, { "id": "870665", "postDate": "06/01/2020 19:51:04", "content": "example of same patient, same mole at different years\n
I think it's right. The main challenge is to propose an approach to handle the evolution of the disease. So, the photos may have taken in different years.
", "rawMarkdown": "I think it's right. The main challenge is to propose an approach to handle the evolution of the disease. So, the photos may have taken in different years.", "votes": null }, { "id": "870955", "postDate": "06/02/2020 04:09:20", "content": "I believe the images are correct and a history for the patient(s). Will try to look at images to get a feel if the camera conditions different or the different years.
\n\nI have many hundreds of these moles - I get checked every few years. About every 5 or so years one or more are removed for analysis. I don't have the images of mine, but my memory is that each check the lighting, etc looks different in the images - different magnification, different cameras, etc. Earliest I remember images being taken is around 20 years ago - before that it was the Mark Two Eyeball.
\n\nNOT likely that the same mole will be imaged on my checks - I have too many. For sure not likely that a malignant mole will have different years - in my case they need to be removed to determine if malignant or benign - I would assume the same for the 500+ in our data, IE, they were removed for confirmation. I have not checked the malignant images yet - plan to look at all 500+ in next couple of days.
\n\nNeed a Doctor to weigh on on this - but IMO the number of images per patient is the useful info - IF the appearnace looks different to the moles. If they all look similar than someone just went for volume. If they look different than that patient is (like me) a high risk skin type. I think I have at least a dozen or more different types from an appearance point of view.
\n\nMoles are a bit like the Microsoft Malware competition - if you never check or update your virus protection than you never find new malware. If only a single image for a patient than they are not high risk - if lots of images over different ages than they are very high risk AND the patient is being checked often - which means detection of a malignant mole much more likely.
", "rawMarkdown": "I believe the images are correct and a history for the patient(s). Will try to look at images to get a feel if the camera conditions different or the different years. \n\nI have many hundreds of these moles - I get checked every few years. About every 5 or so years one or more are removed for analysis. I don't have the images of mine, but my memory is that each check the lighting, etc looks different in the images - different magnification, different cameras, etc. Earliest I remember images being taken is around 20 years ago - before that it was the Mark Two Eyeball.\n\nNOT likely that the same mole will be imaged on my checks - I have too many. For sure not likely that a malignant mole will have different years - in my case they need to be removed to determine if malignant or benign - I would assume the same for the 500+ in our data, IE, they were removed for confirmation. I have not checked the malignant images yet - plan to look at all 500+ in next couple of days.\n\nNeed a Doctor to weigh on on this - but IMO the number of images per patient is the useful info - IF the appearnace looks different to the moles. If they all look similar than someone just went for volume. If they look different than that patient is (like me) a high risk skin type. I think I have at least a dozen or more different types from an appearance point of view.\n\nMoles are a bit like the Microsoft Malware competition - if you never check or update your virus protection than you never find new malware. If only a single image for a patient than they are not high risk - if lots of images over different ages than they are very high risk AND the patient is being checked often - which means detection of a malignant mole much more likely.", "votes": null }, { "id": "871112", "postDate": "06/02/2020 06:38:10", "content": "Okay, that means more images of a patient is a trait for the cancer being a malignant one, right?
\n\nI believe that's why one person shared her approach which said we can determine about 66% malignant cells accurately only with the patient id feature. If I am thinking right, this is the new information which is in the tabular data which is going to be a lot more helpful.
\n\nThanks for sharing the info! :)
", "rawMarkdown": "Okay, that means more images of a patient is a trait for the cancer being a malignant one, right?\n\nI believe that's why one person shared her approach which said we can determine about 66% malignant cells accurately only with the patient id feature. If I am thinking right, this is the new information which is in the tabular data which is going to be a lot more helpful.\n\nThanks for sharing the info! :)", "votes": null }, { "id": "871870", "postDate": "06/02/2020 17:28:08", "content": "The age presented for each image reflects the age of the individual at the time of image capture and is rounded to the nearest multiple of 5. Images for a given patient may not necessarily have been captured at the same appointment, but over the course of multiple years.
", "rawMarkdown": "The age presented for each image reflects the age of the individual at the time of image capture and is rounded to the nearest multiple of 5. Images for a given patient may not necessarily have been captured at the same appointment, but over the course of multiple years.", "votes": null }, { "id": "871948", "postDate": "06/02/2020 18:45:55", "content": "The dataset contains multiple images per patient and was designed to contain exactly one image per lesion, though it is possible that some may have been erroneously attributed as separate lesions.
", "rawMarkdown": "The dataset contains multiple images per patient and was designed to contain exactly one image per lesion, though it is possible that some may have been erroneously attributed as separate lesions.", "votes": null }, { "id": "871954", "postDate": "06/02/2020 18:53:17", "content": "It has been suggested in the literature that the incidence of melanoma is higher among people who have been previously diagnosed with melanoma compared to those with no previous history. However, there is no patient id overlap between the training and testing sets.
", "rawMarkdown": "It has been suggested in the literature that the incidence of melanoma is higher among people who have been previously diagnosed with melanoma compared to those with no previous history. However, there is no patient id overlap between the training and testing sets.", "votes": null }, { "id": "916540", "postDate": "07/05/2020 18:29:06", "content": "Can you please elaborate on how the rounding was performed? I noticed that for many datasets, the age given in the CSV file differs by +- 5 years from the one in the DICOM meta data. So I am assuming, a slightly different method of rounding were used.
", "rawMarkdown": "Can you please elaborate on how the rounding was performed? I noticed that for many datasets, the age given in the CSV file differs by +- 5 years from the one in the DICOM meta data. So I am assuming, a slightly different method of rounding were used.", "votes": null } ], "comments": [ { "id": 870406, "author_name": "maximofn", "author_url": "", "post_date": "06/01/2020 16:58:39", "content": "This interests me
", "votes": null, "replies": [] }, { "id": 870417, "author_name": "maximofn", "author_url": "", "post_date": "06/01/2020 17:04:00", "content": "Is it possible that he has been tested for several years? Or does he have very different ages?
", "votes": null, "replies": [] }, { "id": 870665, "author_name": "hengck23", "author_url": "", "post_date": "06/01/2020 19:51:04", "content": "example of same patient, same mole at different years\n
I think it's right. The main challenge is to propose an approach to handle the evolution of the disease. So, the photos may have taken in different years.
", "votes": null, "replies": [ { "id": 871948, "author_name": "nicholaskurtansky", "author_url": "", "post_date": "06/02/2020 18:45:55", "content": "The dataset contains multiple images per patient and was designed to contain exactly one image per lesion, though it is possible that some may have been erroneously attributed as separate lesions.
", "votes": null, "replies": [] } ] }, { "id": 870955, "author_name": "pcjimmmy", "author_url": "", "post_date": "06/02/2020 04:09:20", "content": "I believe the images are correct and a history for the patient(s). Will try to look at images to get a feel if the camera conditions different or the different years.
\n\nI have many hundreds of these moles - I get checked every few years. About every 5 or so years one or more are removed for analysis. I don't have the images of mine, but my memory is that each check the lighting, etc looks different in the images - different magnification, different cameras, etc. Earliest I remember images being taken is around 20 years ago - before that it was the Mark Two Eyeball.
\n\nNOT likely that the same mole will be imaged on my checks - I have too many. For sure not likely that a malignant mole will have different years - in my case they need to be removed to determine if malignant or benign - I would assume the same for the 500+ in our data, IE, they were removed for confirmation. I have not checked the malignant images yet - plan to look at all 500+ in next couple of days.
\n\nNeed a Doctor to weigh on on this - but IMO the number of images per patient is the useful info - IF the appearnace looks different to the moles. If they all look similar than someone just went for volume. If they look different than that patient is (like me) a high risk skin type. I think I have at least a dozen or more different types from an appearance point of view.
\n\nMoles are a bit like the Microsoft Malware competition - if you never check or update your virus protection than you never find new malware. If only a single image for a patient than they are not high risk - if lots of images over different ages than they are very high risk AND the patient is being checked often - which means detection of a malignant mole much more likely.
", "votes": null, "replies": [ { "id": 871112, "author_name": "sarques", "author_url": "", "post_date": "06/02/2020 06:38:10", "content": "Okay, that means more images of a patient is a trait for the cancer being a malignant one, right?
\n\nI believe that's why one person shared her approach which said we can determine about 66% malignant cells accurately only with the patient id feature. If I am thinking right, this is the new information which is in the tabular data which is going to be a lot more helpful.
\n\nThanks for sharing the info! :)
", "votes": null, "replies": [] }, { "id": 871954, "author_name": "nicholaskurtansky", "author_url": "", "post_date": "06/02/2020 18:53:17", "content": "It has been suggested in the literature that the incidence of melanoma is higher among people who have been previously diagnosed with melanoma compared to those with no previous history. However, there is no patient id overlap between the training and testing sets.
", "votes": null, "replies": [] } ] }, { "id": 871870, "author_name": "nicholaskurtansky", "author_url": "", "post_date": "06/02/2020 17:28:08", "content": "The age presented for each image reflects the age of the individual at the time of image capture and is rounded to the nearest multiple of 5. Images for a given patient may not necessarily have been captured at the same appointment, but over the course of multiple years.
", "votes": null, "replies": [ { "id": 916540, "author_name": "alphanoob1337", "author_url": "", "post_date": "07/05/2020 18:29:06", "content": "Can you please elaborate on how the rounding was performed? I noticed that for many datasets, the age given in the CSV file differs by +- 5 years from the one in the DICOM meta data. So I am assuming, a slightly different method of rounding were used.
", "votes": null, "replies": [] } ] } ], "raw_markdown_by_id": { "870369": "This dataset has many records for the same patient having varied age values. For example, Patient ID IP_0038545 has 28 records, having age values varying from 45 to 60. Shall we take it as erroneous data or the records themselves taken at different ages ?", "870406": "This interests me", "870417": "Is it possible that he has been tested for several years? Or does he have very different ages?", "870665": "example of same patient, same mole at different years\n", "870734": "I think it's right. The main challenge is to propose an approach to handle the evolution of the disease. So, the photos may have taken in different years.", "870955": "I believe the images are correct and a history for the patient(s). Will try to look at images to get a feel if the camera conditions different or the different years. \n\nI have many hundreds of these moles - I get checked every few years. About every 5 or so years one or more are removed for analysis. I don't have the images of mine, but my memory is that each check the lighting, etc looks different in the images - different magnification, different cameras, etc. Earliest I remember images being taken is around 20 years ago - before that it was the Mark Two Eyeball.\n\nNOT likely that the same mole will be imaged on my checks - I have too many. For sure not likely that a malignant mole will have different years - in my case they need to be removed to determine if malignant or benign - I would assume the same for the 500+ in our data, IE, they were removed for confirmation. I have not checked the malignant images yet - plan to look at all 500+ in next couple of days.\n\nNeed a Doctor to weigh on on this - but IMO the number of images per patient is the useful info - IF the appearnace looks different to the moles. If they all look similar than someone just went for volume. If they look different than that patient is (like me) a high risk skin type. I think I have at least a dozen or more different types from an appearance point of view.\n\nMoles are a bit like the Microsoft Malware competition - if you never check or update your virus protection than you never find new malware. If only a single image for a patient than they are not high risk - if lots of images over different ages than they are very high risk AND the patient is being checked often - which means detection of a malignant mole much more likely.", "871112": "Okay, that means more images of a patient is a trait for the cancer being a malignant one, right?\n\nI believe that's why one person shared her approach which said we can determine about 66% malignant cells accurately only with the patient id feature. If I am thinking right, this is the new information which is in the tabular data which is going to be a lot more helpful.\n\nThanks for sharing the info! :)", "871870": "The age presented for each image reflects the age of the individual at the time of image capture and is rounded to the nearest multiple of 5. Images for a given patient may not necessarily have been captured at the same appointment, but over the course of multiple years.", "871948": "The dataset contains multiple images per patient and was designed to contain exactly one image per lesion, though it is possible that some may have been erroneously attributed as separate lesions.", "871954": "It has been suggested in the literature that the incidence of melanoma is higher among people who have been previously diagnosed with melanoma compared to those with no previous history. However, there is no patient id overlap between the training and testing sets.", "916540": "Can you please elaborate on how the rounding was performed? I noticed that for many datasets, the age given in the CSV file differs by +- 5 years from the one in the DICOM meta data. So I am assuming, a slightly different method of rounding were used." }, "source": "meta" }