Datasets:

tequilana
/

OncoEval-MM

Error code:   StreamingRowsError
Exception:    CastError
Message:      Couldn't cast
firstLinesChars: int64
blocks: list<item: struct<text: string>>
  child 0, item: struct<text: string>
      child 0, text: string
markdownContent: string
index: int64
subType: string
lineHeight: int64
text: string
alignment: string
type: string
pageNum: int64
uniqueId: string
level: int64
content: string
structure: list<item: string>
  child 0, item: string
page: int64
to
{'content': Value('string'), 'page': Value('int64'), 'structure': List(Value('string'))}
because column names don't match
Traceback:    Traceback (most recent call last):
                File "/src/services/worker/src/worker/utils.py", line 147, in get_rows_or_raise
                  return get_rows(
                      dataset=dataset,
                  ...<4 lines>...
                      column_names=column_names,
                  )
                File "/src/libs/libcommon/src/libcommon/utils.py", line 272, in decorator
                  return func(*args, **kwargs)
                File "/src/services/worker/src/worker/utils.py", line 127, in get_rows
                  rows_plus_one = list(itertools.islice(safe_iter(ds, dataset=dataset), rows_max_number + 1))
                File "/src/services/worker/src/worker/utils.py", line 483, in safe_iter
                  yield from ds.decode(False) if ds.features else ds
                File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 2840, in __iter__
                  for key, example in ex_iterable:
                                      ^^^^^^^^^^^
                File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 2373, in __iter__
                  for key, pa_table in self._iter_arrow():
                                       ~~~~~~~~~~~~~~~~^^
                File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 2398, in _iter_arrow
                  for key, pa_table in self.ex_iterable._iter_arrow():
                                       ~~~~~~~~~~~~~~~~~~~~~~~~~~~~^^
                File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 536, in _iter_arrow
                  for key, pa_table in iterator:
                                       ^^^^^^^^
                File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 419, in _iter_arrow
                  for key, pa_table in self.generate_tables_fn(**gen_kwags):
                                       ~~~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^
                File "/usr/local/lib/python3.14/site-packages/datasets/packaged_modules/json/json.py", line 343, in _generate_tables
                  self._cast_table(pa_table, json_field_paths=json_field_paths),
                  ~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
                File "/usr/local/lib/python3.14/site-packages/datasets/packaged_modules/json/json.py", line 132, in _cast_table
                  pa_table = table_cast(pa_table, self.info.features.arrow_schema)
                File "/usr/local/lib/python3.14/site-packages/datasets/table.py", line 2378, in table_cast
                  return cast_table_to_schema(table, schema)
                File "/usr/local/lib/python3.14/site-packages/datasets/table.py", line 2306, in cast_table_to_schema
                  raise CastError(
                  ...<3 lines>...
                  )
              datasets.table.CastError: Couldn't cast
              firstLinesChars: int64
              blocks: list<item: struct<text: string>>
                child 0, item: struct<text: string>
                    child 0, text: string
              markdownContent: string
              index: int64
              subType: string
              lineHeight: int64
              text: string
              alignment: string
              type: string
              pageNum: int64
              uniqueId: string
              level: int64
              content: string
              structure: list<item: string>
                child 0, item: string
              page: int64
              to
              {'content': Value('string'), 'page': Value('int64'), 'structure': List(Value('string'))}
              because column names don't match

Need help to make the dataset viewer work? Make sure to review how to configure the dataset viewer, and open a discussion for direct support.

OncoEval-MM

OncoEval-MM is a workflow-structured multimodal dataset for clinical reasoning in oncology. Each case starts from a published narrative report and is stored as an image-centric record: every figure is linked to findings, laboratory results, timeline, and the surrounding clinical context.

The dataset covers eight oncological domains (gastric, pancreatic, prostate, hepatic, colorectal, breast, cervical, and lung cancer) and ten imaging modalities (CT, MRI, PET, ultrasound, X-ray, mammography, endoscopy, histopathology, molecular/genetic testing, and other).

Companion evaluation code is available at lixt099/OncoEval-MM on GitHub, including setup, dataset construction, inference, and evaluation instructions. Set ONCOEVAL_DATA in the code repository to this dataset's data/ directory; the default is ../OncoEval-MM/data when the two repositories are cloned side by side.

Directory layout

data/
  胃癌/                 gastric cancer
  胰腺癌/               pancreatic cancer
  前列腺癌/             prostate cancer
  肝癌/                 hepatic cancer
  结肠癌/               colorectal cancer
  乳腺癌/               breast cancer
  宫颈癌/               cervical cancer
  肺癌/                 lung cancer
    <case name>/
      image/                          PNG figures
      index.json                      image id → path + caption
      <case>.jsonl                    source report text
      <case>_original.json            parsed source record
      <case>_output.json              structured report sections
      analysis_results.json           image-centric findings
      clinical_results.json           pre-diagnosis context
      treatment_results.json          pre-treatment context

Case files

index.json maps each figure id to a relative image_path and a caption.

analysis_results.json is a list, one object per figure, with:

main_findings (summary + details)
body_part_examined
imaging_modality
timeline
clinical_context
diagnostic_inference
image_filename

clinical_results.json holds the Diagnostic Reasoning input: background, lab_results, pre-diagnosis picture entries, and diagnosis_info as the reference diagnosis. Background and picture fields do not contain the final diagnosis.

treatment_results.json holds the Treatment Planning input: background, lab_results, pre-treatment picture entries, and treatment_info as the reference plan.

*_output.json stores the sectioned report (摘要, 一般资料, 检查结果, 诊断与鉴别诊断, 治疗方案, 治疗结果、随访及转归, …). Construction scripts read 治疗方案 from this file.

Tasks supported by these files

Task	Fields used
Medical Image Examination	`analysis_results.json` findings + `image/`
Diagnostic Reasoning	`clinical_results.json` + pre-diagnosis images
Treatment Planning	`treatment_results.json` + pre-treatment images
Medical Attribute Recognition	`body_part_examined`, `imaging_modality`
Anatomical Lesion Localization	image + lesion boxes
Text-to-Image Evidence Matching	findings summary vs candidate images in the same case
Evidence-Aware Reliability	original images and degraded copies produced at evaluation time

License

Released under CC BY 4.0. Source case reports remain subject to their original publication terms.

Downloads last month: 37

Total file size:

1.28 GB