ValueError: On entry to DGELSD parameter number 12 had an illegal value

[Jose Manuel Saborit Torres]

Hi:

  We are creating a function (for T2 mapping computation) that has two arguments: an array of MRI images, and an array of different float values. this is the function:

def find_t2(nifti_images, tes):

    

    data_dicom = [pydicom.dcmread(fn) for fn in nifti_images]

    data_sin_log=np.array([fn.pixel_array for fn in data_dicom]).astype("float32")

    data = np.log(data_sin_log)

    data[data < 0 ] = 0

    tes = np.array(tes)

    x = np.concatenate((np.ones_like(tes[..., np.newaxis]), -tes[..., np.newaxis]), 1)

    

    beta, _, _, _ = sp.linalg.lstsq(x, data)

    s0_ = np.exp(beta[0])

    t2_ = 1./beta[1]

    

    filename="/".join(nifti_images.split("/")[:-2])

    data_s0 = data_dicom[0].copy()

    data_t2 = data_dicom[0].copy()

    data_s0.pixel_array=s0_

    data_t2.pixel_array=t2_

    pydicom.filewriter.dcmwrite(os.path.join(filename,"s0.dcm"), data_s0, write_like_original=True)

    pydicom.filewriter.dcmwrite(os.path.join(filename,"t2.dcm"), data_t2, write_like_original=True)

    return fn_s0, fn_t2

  In the line "beta, _, _, _ = sp.linalg.lstsq(x, data)" we got this error:

---------------------------------------------------------------------------
ValueError                                Traceback (most recent call last)
<ipython-input-99-7c1bf581e9d2> in <module>
      6 
      7 for index in range(0,len(arr),8):
----> 8     find_t2(arr[index:index+8], T_e_array)
      9     break

<ipython-input-98-38eeaf7637e3> in find_t2(nifti_images, tes)
     10     x = np.concatenate((np.ones_like(tes[..., np.newaxis]), -tes[..., np.newaxis]), 1)
     11 
---> 12     beta, _, _, _ = sp.linalg.lstsq(x, data)
     13 
     14     s0_ = np.exp(beta[0])

~/python3.7_venv/lib/python3.7/site-packages/scipy/linalg/basic.py in lstsq(a, b, cond, overwrite_a, overwrite_b, check_finite, lapack_driver)
   1209                                                     overwrite_a=overwrite_a,
   1210                                                     overwrite_b=overwrite_b)
-> 1211 
   1212         elif driver == 'gelsd':
   1213             if real_data:

ValueError: On entry to DGELSD parameter number 12 had an illegal value

The matrix of images we converted in float type and there haven't NaN. The version of scipy is 1.4.1.

How can we solve that? Please answer ASAP.

Sincerely,

Jose Manuel